You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

TransportHttp.java 31KB

Rewrite reference handling to be abstract and accurate This commit actually does three major changes to the way references are handled within JGit. Unfortunately they were easier to do as a single massive commit than to break them up into smaller units. Disambiguate symbolic references: --------------------------------- Reporting a symbolic reference such as HEAD as though it were any other normal reference like refs/heads/master causes subtle programming errors. We have been bitten by this error on several occasions, as have some downstream applications written by myself. Instead of reporting HEAD as a reference whose name differs from its "original name", report it as an actual SymbolicRef object that the application can test the type and examine the target of. With this change, Ref is now an abstract type with different subclasses for the different types. In the classical example of "HEAD" being a symbolic reference to branch "refs/heads/master", the Repository.getAllRefs() method will now return: Map<String, Ref> all = repository.getAllRefs(); SymbolicRef HEAD = (SymbolicRef) all.get("HEAD"); ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master"); assertSame(master, HEAD.getTarget()); assertSame(master.getObjectId(), HEAD.getObjectId()); assertEquals("HEAD", HEAD.getName()); assertEquals("refs/heads/master", master.getName()); A nice side-effect of this change is the storage type of the symbolic reference is no longer ambiguous with the storge type of the underlying reference it targets. In the above example, if master was only available in the packed-refs file, then the following is also true: assertSame(Ref.Storage.LOOSE, HEAD.getStorage()); assertSame(Ref.Storage.PACKED, master.getStorage()); (Prior to this change we returned the ambiguous storage of LOOSE_PACKED for HEAD, which was confusing since it wasn't actually true on disk). Another nice side-effect of this change is all intermediate symbolic references are preserved, and are therefore visible to the application when they walk the target chain. We can now correctly inspect chains of symbolic references. As a result of this change the Ref.getOrigName() method has been removed from the API. Applications should identify a symbolic reference by testing for isSymbolic() and not by using an arcane string comparsion between properties. Abstract the RefDatabase storage: --------------------------------- RefDatabase is now abstract, similar to ObjectDatabase, and a new concrete implementation called RefDirectory is used for the traditional on-disk storage layout. In the future we plan to support additional implementations, such as a pure in-memory RefDatabase for unit testing purposes. Optimize RefDirectory: ---------------------- The implementation of the in-memory reference cache, reading, and update routines has been completely rewritten. Much of the code was heavily borrowed or cribbed from the prior implementation, so copyright notices have been left intact as much as possible. The RefDirectory cache no longer confuses symbolic references with normal references. This permits the cache to resolve the value of a symbolic reference as late as possible, ensuring it is always current, without needing to maintain reverse pointers. The cache is now 2 sorted RefLists, rather than 3 HashMaps. Using sorted lists allows the implementation to reduce the in-memory footprint when storing many refs. Using specialized types for the elements allows the code to avoid additional map lookups for auxiliary stat information. To improve scan time during getRefs(), the lists are returned via a copy-on-write contract. Most callers of getRefs() do not modify the returned collections, so the copy-on-write semantics improves access on repositories with a large number of packed references. Iterator traversals of the returned Map<String,Ref> are performed using a simple merge-join of the two cache lists, ensuring we can perform the entire traversal in linear time as a function of the number of references: O(PackedRefs + LooseRefs). Scans of the loose reference space to update the cache run in O(LooseRefs log LooseRefs) time, as the directory contents are sorted before being merged against the in-memory cache. Since the majority of stable references are kept packed, there typically are only a handful of reference names to be sorted, so the sorting cost should not be very high. Locking is reduced during getRefs() by taking advantage of the copy-on-write semantics of the improved cache data structure. This permits concurrent readers to pull back references without blocking each other. If there is contention updating the cache during a scan, one or more updates are simply skipped and will get picked up again in a future scan. Writing to the $GIT_DIR/packed-refs during reference delete is now fully atomic. The file is locked, reparsed fresh, and written back out if a change is necessary. This avoids all race conditions with concurrent external updates of the packed-refs file. The RefLogWriter class has been fully folded into RefDirectory and is therefore deleted. Maintaining the reference's log is the responsiblity of the database implementation, and not all implementations will use java.io for access. Future work still remains to be done to abstract the ReflogReader class away from local disk IO. Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 years ago
Rewrite reference handling to be abstract and accurate This commit actually does three major changes to the way references are handled within JGit. Unfortunately they were easier to do as a single massive commit than to break them up into smaller units. Disambiguate symbolic references: --------------------------------- Reporting a symbolic reference such as HEAD as though it were any other normal reference like refs/heads/master causes subtle programming errors. We have been bitten by this error on several occasions, as have some downstream applications written by myself. Instead of reporting HEAD as a reference whose name differs from its "original name", report it as an actual SymbolicRef object that the application can test the type and examine the target of. With this change, Ref is now an abstract type with different subclasses for the different types. In the classical example of "HEAD" being a symbolic reference to branch "refs/heads/master", the Repository.getAllRefs() method will now return: Map<String, Ref> all = repository.getAllRefs(); SymbolicRef HEAD = (SymbolicRef) all.get("HEAD"); ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master"); assertSame(master, HEAD.getTarget()); assertSame(master.getObjectId(), HEAD.getObjectId()); assertEquals("HEAD", HEAD.getName()); assertEquals("refs/heads/master", master.getName()); A nice side-effect of this change is the storage type of the symbolic reference is no longer ambiguous with the storge type of the underlying reference it targets. In the above example, if master was only available in the packed-refs file, then the following is also true: assertSame(Ref.Storage.LOOSE, HEAD.getStorage()); assertSame(Ref.Storage.PACKED, master.getStorage()); (Prior to this change we returned the ambiguous storage of LOOSE_PACKED for HEAD, which was confusing since it wasn't actually true on disk). Another nice side-effect of this change is all intermediate symbolic references are preserved, and are therefore visible to the application when they walk the target chain. We can now correctly inspect chains of symbolic references. As a result of this change the Ref.getOrigName() method has been removed from the API. Applications should identify a symbolic reference by testing for isSymbolic() and not by using an arcane string comparsion between properties. Abstract the RefDatabase storage: --------------------------------- RefDatabase is now abstract, similar to ObjectDatabase, and a new concrete implementation called RefDirectory is used for the traditional on-disk storage layout. In the future we plan to support additional implementations, such as a pure in-memory RefDatabase for unit testing purposes. Optimize RefDirectory: ---------------------- The implementation of the in-memory reference cache, reading, and update routines has been completely rewritten. Much of the code was heavily borrowed or cribbed from the prior implementation, so copyright notices have been left intact as much as possible. The RefDirectory cache no longer confuses symbolic references with normal references. This permits the cache to resolve the value of a symbolic reference as late as possible, ensuring it is always current, without needing to maintain reverse pointers. The cache is now 2 sorted RefLists, rather than 3 HashMaps. Using sorted lists allows the implementation to reduce the in-memory footprint when storing many refs. Using specialized types for the elements allows the code to avoid additional map lookups for auxiliary stat information. To improve scan time during getRefs(), the lists are returned via a copy-on-write contract. Most callers of getRefs() do not modify the returned collections, so the copy-on-write semantics improves access on repositories with a large number of packed references. Iterator traversals of the returned Map<String,Ref> are performed using a simple merge-join of the two cache lists, ensuring we can perform the entire traversal in linear time as a function of the number of references: O(PackedRefs + LooseRefs). Scans of the loose reference space to update the cache run in O(LooseRefs log LooseRefs) time, as the directory contents are sorted before being merged against the in-memory cache. Since the majority of stable references are kept packed, there typically are only a handful of reference names to be sorted, so the sorting cost should not be very high. Locking is reduced during getRefs() by taking advantage of the copy-on-write semantics of the improved cache data structure. This permits concurrent readers to pull back references without blocking each other. If there is contention updating the cache during a scan, one or more updates are simply skipped and will get picked up again in a future scan. Writing to the $GIT_DIR/packed-refs during reference delete is now fully atomic. The file is locked, reparsed fresh, and written back out if a change is necessary. This avoids all race conditions with concurrent external updates of the packed-refs file. The RefLogWriter class has been fully folded into RefDirectory and is therefore deleted. Maintaining the reference's log is the responsiblity of the database implementation, and not all implementations will use java.io for access. Future work still remains to be done to abstract the ReflogReader class away from local disk IO. Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 years ago
Rewrite reference handling to be abstract and accurate This commit actually does three major changes to the way references are handled within JGit. Unfortunately they were easier to do as a single massive commit than to break them up into smaller units. Disambiguate symbolic references: --------------------------------- Reporting a symbolic reference such as HEAD as though it were any other normal reference like refs/heads/master causes subtle programming errors. We have been bitten by this error on several occasions, as have some downstream applications written by myself. Instead of reporting HEAD as a reference whose name differs from its "original name", report it as an actual SymbolicRef object that the application can test the type and examine the target of. With this change, Ref is now an abstract type with different subclasses for the different types. In the classical example of "HEAD" being a symbolic reference to branch "refs/heads/master", the Repository.getAllRefs() method will now return: Map<String, Ref> all = repository.getAllRefs(); SymbolicRef HEAD = (SymbolicRef) all.get("HEAD"); ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master"); assertSame(master, HEAD.getTarget()); assertSame(master.getObjectId(), HEAD.getObjectId()); assertEquals("HEAD", HEAD.getName()); assertEquals("refs/heads/master", master.getName()); A nice side-effect of this change is the storage type of the symbolic reference is no longer ambiguous with the storge type of the underlying reference it targets. In the above example, if master was only available in the packed-refs file, then the following is also true: assertSame(Ref.Storage.LOOSE, HEAD.getStorage()); assertSame(Ref.Storage.PACKED, master.getStorage()); (Prior to this change we returned the ambiguous storage of LOOSE_PACKED for HEAD, which was confusing since it wasn't actually true on disk). Another nice side-effect of this change is all intermediate symbolic references are preserved, and are therefore visible to the application when they walk the target chain. We can now correctly inspect chains of symbolic references. As a result of this change the Ref.getOrigName() method has been removed from the API. Applications should identify a symbolic reference by testing for isSymbolic() and not by using an arcane string comparsion between properties. Abstract the RefDatabase storage: --------------------------------- RefDatabase is now abstract, similar to ObjectDatabase, and a new concrete implementation called RefDirectory is used for the traditional on-disk storage layout. In the future we plan to support additional implementations, such as a pure in-memory RefDatabase for unit testing purposes. Optimize RefDirectory: ---------------------- The implementation of the in-memory reference cache, reading, and update routines has been completely rewritten. Much of the code was heavily borrowed or cribbed from the prior implementation, so copyright notices have been left intact as much as possible. The RefDirectory cache no longer confuses symbolic references with normal references. This permits the cache to resolve the value of a symbolic reference as late as possible, ensuring it is always current, without needing to maintain reverse pointers. The cache is now 2 sorted RefLists, rather than 3 HashMaps. Using sorted lists allows the implementation to reduce the in-memory footprint when storing many refs. Using specialized types for the elements allows the code to avoid additional map lookups for auxiliary stat information. To improve scan time during getRefs(), the lists are returned via a copy-on-write contract. Most callers of getRefs() do not modify the returned collections, so the copy-on-write semantics improves access on repositories with a large number of packed references. Iterator traversals of the returned Map<String,Ref> are performed using a simple merge-join of the two cache lists, ensuring we can perform the entire traversal in linear time as a function of the number of references: O(PackedRefs + LooseRefs). Scans of the loose reference space to update the cache run in O(LooseRefs log LooseRefs) time, as the directory contents are sorted before being merged against the in-memory cache. Since the majority of stable references are kept packed, there typically are only a handful of reference names to be sorted, so the sorting cost should not be very high. Locking is reduced during getRefs() by taking advantage of the copy-on-write semantics of the improved cache data structure. This permits concurrent readers to pull back references without blocking each other. If there is contention updating the cache during a scan, one or more updates are simply skipped and will get picked up again in a future scan. Writing to the $GIT_DIR/packed-refs during reference delete is now fully atomic. The file is locked, reparsed fresh, and written back out if a change is necessary. This avoids all race conditions with concurrent external updates of the packed-refs file. The RefLogWriter class has been fully folded into RefDirectory and is therefore deleted. Maintaining the reference's log is the responsiblity of the database implementation, and not all implementations will use java.io for access. Future work still remains to be done to abstract the ReflogReader class away from local disk IO. Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 years ago
Rewrite reference handling to be abstract and accurate This commit actually does three major changes to the way references are handled within JGit. Unfortunately they were easier to do as a single massive commit than to break them up into smaller units. Disambiguate symbolic references: --------------------------------- Reporting a symbolic reference such as HEAD as though it were any other normal reference like refs/heads/master causes subtle programming errors. We have been bitten by this error on several occasions, as have some downstream applications written by myself. Instead of reporting HEAD as a reference whose name differs from its "original name", report it as an actual SymbolicRef object that the application can test the type and examine the target of. With this change, Ref is now an abstract type with different subclasses for the different types. In the classical example of "HEAD" being a symbolic reference to branch "refs/heads/master", the Repository.getAllRefs() method will now return: Map<String, Ref> all = repository.getAllRefs(); SymbolicRef HEAD = (SymbolicRef) all.get("HEAD"); ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master"); assertSame(master, HEAD.getTarget()); assertSame(master.getObjectId(), HEAD.getObjectId()); assertEquals("HEAD", HEAD.getName()); assertEquals("refs/heads/master", master.getName()); A nice side-effect of this change is the storage type of the symbolic reference is no longer ambiguous with the storge type of the underlying reference it targets. In the above example, if master was only available in the packed-refs file, then the following is also true: assertSame(Ref.Storage.LOOSE, HEAD.getStorage()); assertSame(Ref.Storage.PACKED, master.getStorage()); (Prior to this change we returned the ambiguous storage of LOOSE_PACKED for HEAD, which was confusing since it wasn't actually true on disk). Another nice side-effect of this change is all intermediate symbolic references are preserved, and are therefore visible to the application when they walk the target chain. We can now correctly inspect chains of symbolic references. As a result of this change the Ref.getOrigName() method has been removed from the API. Applications should identify a symbolic reference by testing for isSymbolic() and not by using an arcane string comparsion between properties. Abstract the RefDatabase storage: --------------------------------- RefDatabase is now abstract, similar to ObjectDatabase, and a new concrete implementation called RefDirectory is used for the traditional on-disk storage layout. In the future we plan to support additional implementations, such as a pure in-memory RefDatabase for unit testing purposes. Optimize RefDirectory: ---------------------- The implementation of the in-memory reference cache, reading, and update routines has been completely rewritten. Much of the code was heavily borrowed or cribbed from the prior implementation, so copyright notices have been left intact as much as possible. The RefDirectory cache no longer confuses symbolic references with normal references. This permits the cache to resolve the value of a symbolic reference as late as possible, ensuring it is always current, without needing to maintain reverse pointers. The cache is now 2 sorted RefLists, rather than 3 HashMaps. Using sorted lists allows the implementation to reduce the in-memory footprint when storing many refs. Using specialized types for the elements allows the code to avoid additional map lookups for auxiliary stat information. To improve scan time during getRefs(), the lists are returned via a copy-on-write contract. Most callers of getRefs() do not modify the returned collections, so the copy-on-write semantics improves access on repositories with a large number of packed references. Iterator traversals of the returned Map<String,Ref> are performed using a simple merge-join of the two cache lists, ensuring we can perform the entire traversal in linear time as a function of the number of references: O(PackedRefs + LooseRefs). Scans of the loose reference space to update the cache run in O(LooseRefs log LooseRefs) time, as the directory contents are sorted before being merged against the in-memory cache. Since the majority of stable references are kept packed, there typically are only a handful of reference names to be sorted, so the sorting cost should not be very high. Locking is reduced during getRefs() by taking advantage of the copy-on-write semantics of the improved cache data structure. This permits concurrent readers to pull back references without blocking each other. If there is contention updating the cache during a scan, one or more updates are simply skipped and will get picked up again in a future scan. Writing to the $GIT_DIR/packed-refs during reference delete is now fully atomic. The file is locked, reparsed fresh, and written back out if a change is necessary. This avoids all race conditions with concurrent external updates of the packed-refs file. The RefLogWriter class has been fully folded into RefDirectory and is therefore deleted. Maintaining the reference's log is the responsiblity of the database implementation, and not all implementations will use java.io for access. Future work still remains to be done to abstract the ReflogReader class away from local disk IO. Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 years ago
Rewrite reference handling to be abstract and accurate This commit actually does three major changes to the way references are handled within JGit. Unfortunately they were easier to do as a single massive commit than to break them up into smaller units. Disambiguate symbolic references: --------------------------------- Reporting a symbolic reference such as HEAD as though it were any other normal reference like refs/heads/master causes subtle programming errors. We have been bitten by this error on several occasions, as have some downstream applications written by myself. Instead of reporting HEAD as a reference whose name differs from its "original name", report it as an actual SymbolicRef object that the application can test the type and examine the target of. With this change, Ref is now an abstract type with different subclasses for the different types. In the classical example of "HEAD" being a symbolic reference to branch "refs/heads/master", the Repository.getAllRefs() method will now return: Map<String, Ref> all = repository.getAllRefs(); SymbolicRef HEAD = (SymbolicRef) all.get("HEAD"); ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master"); assertSame(master, HEAD.getTarget()); assertSame(master.getObjectId(), HEAD.getObjectId()); assertEquals("HEAD", HEAD.getName()); assertEquals("refs/heads/master", master.getName()); A nice side-effect of this change is the storage type of the symbolic reference is no longer ambiguous with the storge type of the underlying reference it targets. In the above example, if master was only available in the packed-refs file, then the following is also true: assertSame(Ref.Storage.LOOSE, HEAD.getStorage()); assertSame(Ref.Storage.PACKED, master.getStorage()); (Prior to this change we returned the ambiguous storage of LOOSE_PACKED for HEAD, which was confusing since it wasn't actually true on disk). Another nice side-effect of this change is all intermediate symbolic references are preserved, and are therefore visible to the application when they walk the target chain. We can now correctly inspect chains of symbolic references. As a result of this change the Ref.getOrigName() method has been removed from the API. Applications should identify a symbolic reference by testing for isSymbolic() and not by using an arcane string comparsion between properties. Abstract the RefDatabase storage: --------------------------------- RefDatabase is now abstract, similar to ObjectDatabase, and a new concrete implementation called RefDirectory is used for the traditional on-disk storage layout. In the future we plan to support additional implementations, such as a pure in-memory RefDatabase for unit testing purposes. Optimize RefDirectory: ---------------------- The implementation of the in-memory reference cache, reading, and update routines has been completely rewritten. Much of the code was heavily borrowed or cribbed from the prior implementation, so copyright notices have been left intact as much as possible. The RefDirectory cache no longer confuses symbolic references with normal references. This permits the cache to resolve the value of a symbolic reference as late as possible, ensuring it is always current, without needing to maintain reverse pointers. The cache is now 2 sorted RefLists, rather than 3 HashMaps. Using sorted lists allows the implementation to reduce the in-memory footprint when storing many refs. Using specialized types for the elements allows the code to avoid additional map lookups for auxiliary stat information. To improve scan time during getRefs(), the lists are returned via a copy-on-write contract. Most callers of getRefs() do not modify the returned collections, so the copy-on-write semantics improves access on repositories with a large number of packed references. Iterator traversals of the returned Map<String,Ref> are performed using a simple merge-join of the two cache lists, ensuring we can perform the entire traversal in linear time as a function of the number of references: O(PackedRefs + LooseRefs). Scans of the loose reference space to update the cache run in O(LooseRefs log LooseRefs) time, as the directory contents are sorted before being merged against the in-memory cache. Since the majority of stable references are kept packed, there typically are only a handful of reference names to be sorted, so the sorting cost should not be very high. Locking is reduced during getRefs() by taking advantage of the copy-on-write semantics of the improved cache data structure. This permits concurrent readers to pull back references without blocking each other. If there is contention updating the cache during a scan, one or more updates are simply skipped and will get picked up again in a future scan. Writing to the $GIT_DIR/packed-refs during reference delete is now fully atomic. The file is locked, reparsed fresh, and written back out if a change is necessary. This avoids all race conditions with concurrent external updates of the packed-refs file. The RefLogWriter class has been fully folded into RefDirectory and is therefore deleted. Maintaining the reference's log is the responsiblity of the database implementation, and not all implementations will use java.io for access. Future work still remains to be done to abstract the ReflogReader class away from local disk IO. Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 years ago
Rewrite reference handling to be abstract and accurate This commit actually does three major changes to the way references are handled within JGit. Unfortunately they were easier to do as a single massive commit than to break them up into smaller units. Disambiguate symbolic references: --------------------------------- Reporting a symbolic reference such as HEAD as though it were any other normal reference like refs/heads/master causes subtle programming errors. We have been bitten by this error on several occasions, as have some downstream applications written by myself. Instead of reporting HEAD as a reference whose name differs from its "original name", report it as an actual SymbolicRef object that the application can test the type and examine the target of. With this change, Ref is now an abstract type with different subclasses for the different types. In the classical example of "HEAD" being a symbolic reference to branch "refs/heads/master", the Repository.getAllRefs() method will now return: Map<String, Ref> all = repository.getAllRefs(); SymbolicRef HEAD = (SymbolicRef) all.get("HEAD"); ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master"); assertSame(master, HEAD.getTarget()); assertSame(master.getObjectId(), HEAD.getObjectId()); assertEquals("HEAD", HEAD.getName()); assertEquals("refs/heads/master", master.getName()); A nice side-effect of this change is the storage type of the symbolic reference is no longer ambiguous with the storge type of the underlying reference it targets. In the above example, if master was only available in the packed-refs file, then the following is also true: assertSame(Ref.Storage.LOOSE, HEAD.getStorage()); assertSame(Ref.Storage.PACKED, master.getStorage()); (Prior to this change we returned the ambiguous storage of LOOSE_PACKED for HEAD, which was confusing since it wasn't actually true on disk). Another nice side-effect of this change is all intermediate symbolic references are preserved, and are therefore visible to the application when they walk the target chain. We can now correctly inspect chains of symbolic references. As a result of this change the Ref.getOrigName() method has been removed from the API. Applications should identify a symbolic reference by testing for isSymbolic() and not by using an arcane string comparsion between properties. Abstract the RefDatabase storage: --------------------------------- RefDatabase is now abstract, similar to ObjectDatabase, and a new concrete implementation called RefDirectory is used for the traditional on-disk storage layout. In the future we plan to support additional implementations, such as a pure in-memory RefDatabase for unit testing purposes. Optimize RefDirectory: ---------------------- The implementation of the in-memory reference cache, reading, and update routines has been completely rewritten. Much of the code was heavily borrowed or cribbed from the prior implementation, so copyright notices have been left intact as much as possible. The RefDirectory cache no longer confuses symbolic references with normal references. This permits the cache to resolve the value of a symbolic reference as late as possible, ensuring it is always current, without needing to maintain reverse pointers. The cache is now 2 sorted RefLists, rather than 3 HashMaps. Using sorted lists allows the implementation to reduce the in-memory footprint when storing many refs. Using specialized types for the elements allows the code to avoid additional map lookups for auxiliary stat information. To improve scan time during getRefs(), the lists are returned via a copy-on-write contract. Most callers of getRefs() do not modify the returned collections, so the copy-on-write semantics improves access on repositories with a large number of packed references. Iterator traversals of the returned Map<String,Ref> are performed using a simple merge-join of the two cache lists, ensuring we can perform the entire traversal in linear time as a function of the number of references: O(PackedRefs + LooseRefs). Scans of the loose reference space to update the cache run in O(LooseRefs log LooseRefs) time, as the directory contents are sorted before being merged against the in-memory cache. Since the majority of stable references are kept packed, there typically are only a handful of reference names to be sorted, so the sorting cost should not be very high. Locking is reduced during getRefs() by taking advantage of the copy-on-write semantics of the improved cache data structure. This permits concurrent readers to pull back references without blocking each other. If there is contention updating the cache during a scan, one or more updates are simply skipped and will get picked up again in a future scan. Writing to the $GIT_DIR/packed-refs during reference delete is now fully atomic. The file is locked, reparsed fresh, and written back out if a change is necessary. This avoids all race conditions with concurrent external updates of the packed-refs file. The RefLogWriter class has been fully folded into RefDirectory and is therefore deleted. Maintaining the reference's log is the responsiblity of the database implementation, and not all implementations will use java.io for access. Future work still remains to be done to abstract the ReflogReader class away from local disk IO. Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 years ago
1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033
  1. /*
  2. * Copyright (C) 2008-2010, Google Inc.
  3. * Copyright (C) 2008, Shawn O. Pearce <spearce@spearce.org>
  4. * Copyright (C) 2013, Matthias Sohn <matthias.sohn@sap.com>
  5. * and other copyright owners as documented in the project's IP log.
  6. *
  7. * This program and the accompanying materials are made available
  8. * under the terms of the Eclipse Distribution License v1.0 which
  9. * accompanies this distribution, is reproduced below, and is
  10. * available at http://www.eclipse.org/org/documents/edl-v10.php
  11. *
  12. * All rights reserved.
  13. *
  14. * Redistribution and use in source and binary forms, with or
  15. * without modification, are permitted provided that the following
  16. * conditions are met:
  17. *
  18. * - Redistributions of source code must retain the above copyright
  19. * notice, this list of conditions and the following disclaimer.
  20. *
  21. * - Redistributions in binary form must reproduce the above
  22. * copyright notice, this list of conditions and the following
  23. * disclaimer in the documentation and/or other materials provided
  24. * with the distribution.
  25. *
  26. * - Neither the name of the Eclipse Foundation, Inc. nor the
  27. * names of its contributors may be used to endorse or promote
  28. * products derived from this software without specific prior
  29. * written permission.
  30. *
  31. * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
  32. * CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
  33. * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
  34. * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  35. * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
  36. * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  37. * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
  38. * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  39. * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
  40. * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
  41. * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
  42. * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
  43. * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  44. */
  45. package org.eclipse.jgit.transport;
  46. import static org.eclipse.jgit.util.HttpSupport.ENCODING_GZIP;
  47. import static org.eclipse.jgit.util.HttpSupport.HDR_ACCEPT;
  48. import static org.eclipse.jgit.util.HttpSupport.HDR_ACCEPT_ENCODING;
  49. import static org.eclipse.jgit.util.HttpSupport.HDR_CONTENT_ENCODING;
  50. import static org.eclipse.jgit.util.HttpSupport.HDR_CONTENT_TYPE;
  51. import static org.eclipse.jgit.util.HttpSupport.HDR_PRAGMA;
  52. import static org.eclipse.jgit.util.HttpSupport.HDR_USER_AGENT;
  53. import static org.eclipse.jgit.util.HttpSupport.HDR_WWW_AUTHENTICATE;
  54. import static org.eclipse.jgit.util.HttpSupport.METHOD_GET;
  55. import static org.eclipse.jgit.util.HttpSupport.METHOD_POST;
  56. import java.io.BufferedReader;
  57. import java.io.ByteArrayInputStream;
  58. import java.io.FileNotFoundException;
  59. import java.io.IOException;
  60. import java.io.InputStream;
  61. import java.io.InputStreamReader;
  62. import java.io.OutputStream;
  63. import java.net.MalformedURLException;
  64. import java.net.Proxy;
  65. import java.net.ProxySelector;
  66. import java.net.URL;
  67. import java.security.KeyManagementException;
  68. import java.security.NoSuchAlgorithmException;
  69. import java.security.cert.X509Certificate;
  70. import java.text.MessageFormat;
  71. import java.util.ArrayList;
  72. import java.util.Arrays;
  73. import java.util.Collection;
  74. import java.util.Collections;
  75. import java.util.EnumSet;
  76. import java.util.LinkedHashSet;
  77. import java.util.Map;
  78. import java.util.Set;
  79. import java.util.TreeMap;
  80. import java.util.zip.GZIPInputStream;
  81. import java.util.zip.GZIPOutputStream;
  82. import javax.net.ssl.HostnameVerifier;
  83. import javax.net.ssl.SSLSession;
  84. import javax.net.ssl.TrustManager;
  85. import javax.net.ssl.X509TrustManager;
  86. import org.eclipse.jgit.errors.NoRemoteRepositoryException;
  87. import org.eclipse.jgit.errors.NotSupportedException;
  88. import org.eclipse.jgit.errors.PackProtocolException;
  89. import org.eclipse.jgit.errors.TransportException;
  90. import org.eclipse.jgit.internal.JGitText;
  91. import org.eclipse.jgit.internal.storage.file.RefDirectory;
  92. import org.eclipse.jgit.lib.Config;
  93. import org.eclipse.jgit.lib.Config.SectionParser;
  94. import org.eclipse.jgit.lib.Constants;
  95. import org.eclipse.jgit.lib.ObjectId;
  96. import org.eclipse.jgit.lib.ObjectIdRef;
  97. import org.eclipse.jgit.lib.ProgressMonitor;
  98. import org.eclipse.jgit.lib.Ref;
  99. import org.eclipse.jgit.lib.Repository;
  100. import org.eclipse.jgit.lib.SymbolicRef;
  101. import org.eclipse.jgit.transport.http.HttpConnection;
  102. import org.eclipse.jgit.util.HttpSupport;
  103. import org.eclipse.jgit.util.IO;
  104. import org.eclipse.jgit.util.RawParseUtils;
  105. import org.eclipse.jgit.util.TemporaryBuffer;
  106. import org.eclipse.jgit.util.io.DisabledOutputStream;
  107. import org.eclipse.jgit.util.io.UnionInputStream;
  108. /**
  109. * Transport over HTTP and FTP protocols.
  110. * <p>
  111. * If the transport is using HTTP and the remote HTTP service is Git-aware
  112. * (speaks the "smart-http protocol") this client will automatically take
  113. * advantage of the additional Git-specific HTTP extensions. If the remote
  114. * service does not support these extensions, the client will degrade to direct
  115. * file fetching.
  116. * <p>
  117. * If the remote (server side) repository does not have the specialized Git
  118. * support, object files are retrieved directly through standard HTTP GET (or
  119. * binary FTP GET) requests. This make it easy to serve a Git repository through
  120. * a standard web host provider that does not offer specific support for Git.
  121. *
  122. * @see WalkFetchConnection
  123. */
  124. public class TransportHttp extends HttpTransport implements WalkTransport,
  125. PackTransport {
  126. private static final String SVC_UPLOAD_PACK = "git-upload-pack"; //$NON-NLS-1$
  127. private static final String SVC_RECEIVE_PACK = "git-receive-pack"; //$NON-NLS-1$
  128. private static final String userAgent = computeUserAgent();
  129. static final TransportProtocol PROTO_HTTP = new TransportProtocol() {
  130. private final String[] schemeNames = { "http", "https" }; //$NON-NLS-1$ //$NON-NLS-2$
  131. private final Set<String> schemeSet = Collections
  132. .unmodifiableSet(new LinkedHashSet<String>(Arrays
  133. .asList(schemeNames)));
  134. public String getName() {
  135. return JGitText.get().transportProtoHTTP;
  136. }
  137. public Set<String> getSchemes() {
  138. return schemeSet;
  139. }
  140. public Set<URIishField> getRequiredFields() {
  141. return Collections.unmodifiableSet(EnumSet.of(URIishField.HOST,
  142. URIishField.PATH));
  143. }
  144. public Set<URIishField> getOptionalFields() {
  145. return Collections.unmodifiableSet(EnumSet.of(URIishField.USER,
  146. URIishField.PASS, URIishField.PORT));
  147. }
  148. public int getDefaultPort() {
  149. return 80;
  150. }
  151. public Transport open(URIish uri, Repository local, String remoteName)
  152. throws NotSupportedException {
  153. return new TransportHttp(local, uri);
  154. }
  155. public Transport open(URIish uri) throws NotSupportedException {
  156. return new TransportHttp(uri);
  157. }
  158. };
  159. static final TransportProtocol PROTO_FTP = new TransportProtocol() {
  160. public String getName() {
  161. return JGitText.get().transportProtoFTP;
  162. }
  163. public Set<String> getSchemes() {
  164. return Collections.singleton("ftp"); //$NON-NLS-1$
  165. }
  166. public Set<URIishField> getRequiredFields() {
  167. return Collections.unmodifiableSet(EnumSet.of(URIishField.HOST,
  168. URIishField.PATH));
  169. }
  170. public Set<URIishField> getOptionalFields() {
  171. return Collections.unmodifiableSet(EnumSet.of(URIishField.USER,
  172. URIishField.PASS, URIishField.PORT));
  173. }
  174. public int getDefaultPort() {
  175. return 21;
  176. }
  177. public Transport open(URIish uri, Repository local, String remoteName)
  178. throws NotSupportedException {
  179. return new TransportHttp(local, uri);
  180. }
  181. };
  182. private static String computeUserAgent() {
  183. String version;
  184. final Package pkg = TransportHttp.class.getPackage();
  185. if (pkg != null && pkg.getImplementationVersion() != null) {
  186. version = pkg.getImplementationVersion();
  187. } else {
  188. version = "unknown"; //$NON-NLS-1$
  189. }
  190. return "JGit/" + version; //$NON-NLS-1$
  191. }
  192. private static final Config.SectionParser<HttpConfig> HTTP_KEY = new SectionParser<HttpConfig>() {
  193. public HttpConfig parse(final Config cfg) {
  194. return new HttpConfig(cfg);
  195. }
  196. };
  197. private static class HttpConfig {
  198. final int postBuffer;
  199. final boolean sslVerify;
  200. HttpConfig(final Config rc) {
  201. postBuffer = rc.getInt("http", "postbuffer", 1 * 1024 * 1024); //$NON-NLS-1$ //$NON-NLS-2$
  202. sslVerify = rc.getBoolean("http", "sslVerify", true); //$NON-NLS-1$ //$NON-NLS-2$
  203. }
  204. private HttpConfig() {
  205. this(new Config());
  206. }
  207. }
  208. private final URL baseUrl;
  209. private final URL objectsUrl;
  210. private final HttpConfig http;
  211. private final ProxySelector proxySelector;
  212. private boolean useSmartHttp = true;
  213. private HttpAuthMethod authMethod = HttpAuthMethod.Type.NONE.method(null);
  214. private Map<String, String> headers;
  215. TransportHttp(final Repository local, final URIish uri)
  216. throws NotSupportedException {
  217. super(local, uri);
  218. try {
  219. String uriString = uri.toString();
  220. if (!uriString.endsWith("/")) //$NON-NLS-1$
  221. uriString += "/"; //$NON-NLS-1$
  222. baseUrl = new URL(uriString);
  223. objectsUrl = new URL(baseUrl, "objects/"); //$NON-NLS-1$
  224. } catch (MalformedURLException e) {
  225. throw new NotSupportedException(MessageFormat.format(JGitText.get().invalidURL, uri), e);
  226. }
  227. http = local.getConfig().get(HTTP_KEY);
  228. proxySelector = ProxySelector.getDefault();
  229. if (getCredentialsProvider() == null)
  230. setCredentialsProvider(new NetRCCredentialsProvider());
  231. }
  232. /**
  233. * Create a minimal HTTP transport with default configuration values.
  234. *
  235. * @param uri
  236. * @throws NotSupportedException
  237. */
  238. TransportHttp(final URIish uri) throws NotSupportedException {
  239. super(uri);
  240. try {
  241. String uriString = uri.toString();
  242. if (!uriString.endsWith("/")) //$NON-NLS-1$
  243. uriString += "/"; //$NON-NLS-1$
  244. baseUrl = new URL(uriString);
  245. objectsUrl = new URL(baseUrl, "objects/"); //$NON-NLS-1$
  246. } catch (MalformedURLException e) {
  247. throw new NotSupportedException(MessageFormat.format(JGitText.get().invalidURL, uri), e);
  248. }
  249. http = new HttpConfig();
  250. proxySelector = ProxySelector.getDefault();
  251. }
  252. /**
  253. * Toggle whether or not smart HTTP transport should be used.
  254. * <p>
  255. * This flag exists primarily to support backwards compatibility testing
  256. * within a testing framework, there is no need to modify it in most
  257. * applications.
  258. *
  259. * @param on
  260. * if {@code true} (default), smart HTTP is enabled.
  261. */
  262. public void setUseSmartHttp(final boolean on) {
  263. useSmartHttp = on;
  264. }
  265. @Override
  266. public FetchConnection openFetch() throws TransportException,
  267. NotSupportedException {
  268. final String service = SVC_UPLOAD_PACK;
  269. try {
  270. final HttpConnection c = connect(service);
  271. final InputStream in = openInputStream(c);
  272. try {
  273. if (isSmartHttp(c, service)) {
  274. readSmartHeaders(in, service);
  275. return new SmartHttpFetchConnection(in);
  276. } else {
  277. // Assume this server doesn't support smart HTTP fetch
  278. // and fall back on dumb object walking.
  279. //
  280. return newDumbConnection(in);
  281. }
  282. } finally {
  283. in.close();
  284. }
  285. } catch (NotSupportedException err) {
  286. throw err;
  287. } catch (TransportException err) {
  288. throw err;
  289. } catch (IOException err) {
  290. throw new TransportException(uri, JGitText.get().errorReadingInfoRefs, err);
  291. }
  292. }
  293. private FetchConnection newDumbConnection(InputStream in)
  294. throws IOException, PackProtocolException {
  295. HttpObjectDB d = new HttpObjectDB(objectsUrl);
  296. BufferedReader br = toBufferedReader(in);
  297. Map<String, Ref> refs;
  298. try {
  299. refs = d.readAdvertisedImpl(br);
  300. } finally {
  301. br.close();
  302. }
  303. if (!refs.containsKey(Constants.HEAD)) {
  304. // If HEAD was not published in the info/refs file (it usually
  305. // is not there) download HEAD by itself as a loose file and do
  306. // the resolution by hand.
  307. //
  308. HttpConnection conn = httpOpen(new URL(baseUrl, Constants.HEAD));
  309. int status = HttpSupport.response(conn);
  310. switch (status) {
  311. case HttpConnection.HTTP_OK: {
  312. br = toBufferedReader(openInputStream(conn));
  313. try {
  314. String line = br.readLine();
  315. if (line != null && line.startsWith(RefDirectory.SYMREF)) {
  316. String target = line.substring(RefDirectory.SYMREF.length());
  317. Ref r = refs.get(target);
  318. if (r == null)
  319. r = new ObjectIdRef.Unpeeled(Ref.Storage.NEW, target, null);
  320. r = new SymbolicRef(Constants.HEAD, r);
  321. refs.put(r.getName(), r);
  322. } else if (line != null && ObjectId.isId(line)) {
  323. Ref r = new ObjectIdRef.Unpeeled(Ref.Storage.NETWORK,
  324. Constants.HEAD, ObjectId.fromString(line));
  325. refs.put(r.getName(), r);
  326. }
  327. } finally {
  328. br.close();
  329. }
  330. break;
  331. }
  332. case HttpConnection.HTTP_NOT_FOUND:
  333. break;
  334. default:
  335. throw new TransportException(uri, MessageFormat.format(
  336. JGitText.get().cannotReadHEAD, Integer.valueOf(status),
  337. conn.getResponseMessage()));
  338. }
  339. }
  340. WalkFetchConnection wfc = new WalkFetchConnection(this, d);
  341. wfc.available(refs);
  342. return wfc;
  343. }
  344. private BufferedReader toBufferedReader(InputStream in) {
  345. return new BufferedReader(new InputStreamReader(in, Constants.CHARSET));
  346. }
  347. @Override
  348. public PushConnection openPush() throws NotSupportedException,
  349. TransportException {
  350. final String service = SVC_RECEIVE_PACK;
  351. try {
  352. final HttpConnection c = connect(service);
  353. final InputStream in = openInputStream(c);
  354. try {
  355. if (isSmartHttp(c, service)) {
  356. readSmartHeaders(in, service);
  357. return new SmartHttpPushConnection(in);
  358. } else if (!useSmartHttp) {
  359. final String msg = JGitText.get().smartHTTPPushDisabled;
  360. throw new NotSupportedException(msg);
  361. } else {
  362. final String msg = JGitText.get().remoteDoesNotSupportSmartHTTPPush;
  363. throw new NotSupportedException(msg);
  364. }
  365. } finally {
  366. in.close();
  367. }
  368. } catch (NotSupportedException err) {
  369. throw err;
  370. } catch (TransportException err) {
  371. throw err;
  372. } catch (IOException err) {
  373. throw new TransportException(uri, JGitText.get().errorReadingInfoRefs, err);
  374. }
  375. }
  376. @Override
  377. public void close() {
  378. // No explicit connections are maintained.
  379. }
  380. /**
  381. * Set additional headers on the HTTP connection
  382. *
  383. * @param headers
  384. * a map of name:values that are to be set as headers on the HTTP
  385. * connection
  386. * @since 3.4
  387. */
  388. public void setAdditionalHeaders(Map<String, String> headers) {
  389. this.headers = headers;
  390. }
  391. private HttpConnection connect(final String service)
  392. throws TransportException, NotSupportedException {
  393. final URL u;
  394. try {
  395. final StringBuilder b = new StringBuilder();
  396. b.append(baseUrl);
  397. if (b.charAt(b.length() - 1) != '/')
  398. b.append('/');
  399. b.append(Constants.INFO_REFS);
  400. if (useSmartHttp) {
  401. b.append(b.indexOf("?") < 0 ? '?' : '&'); //$NON-NLS-1$
  402. b.append("service="); //$NON-NLS-1$
  403. b.append(service);
  404. }
  405. u = new URL(b.toString());
  406. } catch (MalformedURLException e) {
  407. throw new NotSupportedException(MessageFormat.format(JGitText.get().invalidURL, uri), e);
  408. }
  409. try {
  410. int authAttempts = 1;
  411. for (;;) {
  412. final HttpConnection conn = httpOpen(u);
  413. if (useSmartHttp) {
  414. String exp = "application/x-" + service + "-advertisement"; //$NON-NLS-1$ //$NON-NLS-2$
  415. conn.setRequestProperty(HDR_ACCEPT, exp + ", */*"); //$NON-NLS-1$
  416. } else {
  417. conn.setRequestProperty(HDR_ACCEPT, "*/*"); //$NON-NLS-1$
  418. }
  419. final int status = HttpSupport.response(conn);
  420. switch (status) {
  421. case HttpConnection.HTTP_OK:
  422. // Check if HttpConnection did some authentication in the
  423. // background (e.g Kerberos/SPNEGO).
  424. // That may not work for streaming requests and jgit
  425. // explicit authentication would be required
  426. if (authMethod.getType() == HttpAuthMethod.Type.NONE
  427. && conn.getHeaderField(HDR_WWW_AUTHENTICATE) != null)
  428. authMethod = HttpAuthMethod.scanResponse(conn);
  429. return conn;
  430. case HttpConnection.HTTP_NOT_FOUND:
  431. throw new NoRemoteRepositoryException(uri,
  432. MessageFormat.format(JGitText.get().uriNotFound, u));
  433. case HttpConnection.HTTP_UNAUTHORIZED:
  434. authMethod = HttpAuthMethod.scanResponse(conn);
  435. if (authMethod.getType() == HttpAuthMethod.Type.NONE)
  436. throw new TransportException(uri, MessageFormat.format(
  437. JGitText.get().authenticationNotSupported, uri));
  438. CredentialsProvider credentialsProvider = getCredentialsProvider();
  439. if (credentialsProvider == null)
  440. throw new TransportException(uri,
  441. JGitText.get().noCredentialsProvider);
  442. if (authAttempts > 1)
  443. credentialsProvider.reset(uri);
  444. if (3 < authAttempts
  445. || !authMethod.authorize(uri, credentialsProvider)) {
  446. throw new TransportException(uri,
  447. JGitText.get().notAuthorized);
  448. }
  449. authAttempts++;
  450. continue;
  451. case HttpConnection.HTTP_FORBIDDEN:
  452. throw new TransportException(uri, MessageFormat.format(
  453. JGitText.get().serviceNotPermitted, service));
  454. default:
  455. String err = status + " " + conn.getResponseMessage(); //$NON-NLS-1$
  456. throw new TransportException(uri, err);
  457. }
  458. }
  459. } catch (NotSupportedException e) {
  460. throw e;
  461. } catch (TransportException e) {
  462. throw e;
  463. } catch (IOException e) {
  464. throw new TransportException(uri, MessageFormat.format(JGitText.get().cannotOpenService, service), e);
  465. }
  466. }
  467. final HttpConnection httpOpen(URL u) throws IOException {
  468. return httpOpen(METHOD_GET, u);
  469. }
  470. /**
  471. * Open an HTTP connection.
  472. *
  473. * @param method
  474. * @param u
  475. * @return the connection
  476. * @throws IOException
  477. * @since 3.3
  478. */
  479. protected HttpConnection httpOpen(String method, URL u)
  480. throws IOException {
  481. final Proxy proxy = HttpSupport.proxyFor(proxySelector, u);
  482. HttpConnection conn = connectionFactory.create(u, proxy);
  483. if (!http.sslVerify && "https".equals(u.getProtocol())) { //$NON-NLS-1$
  484. disableSslVerify(conn);
  485. }
  486. conn.setRequestMethod(method);
  487. conn.setUseCaches(false);
  488. conn.setRequestProperty(HDR_ACCEPT_ENCODING, ENCODING_GZIP);
  489. conn.setRequestProperty(HDR_PRAGMA, "no-cache"); //$NON-NLS-1$
  490. conn.setRequestProperty(HDR_USER_AGENT, userAgent);
  491. int timeOut = getTimeout();
  492. if (timeOut != -1) {
  493. int effTimeOut = timeOut * 1000;
  494. conn.setConnectTimeout(effTimeOut);
  495. conn.setReadTimeout(effTimeOut);
  496. }
  497. if (this.headers != null && !this.headers.isEmpty()) {
  498. for (Map.Entry<String, String> entry : this.headers.entrySet())
  499. conn.setRequestProperty(entry.getKey(), entry.getValue());
  500. }
  501. authMethod.configureRequest(conn);
  502. return conn;
  503. }
  504. private void disableSslVerify(HttpConnection conn)
  505. throws IOException {
  506. final TrustManager[] trustAllCerts = new TrustManager[] { new DummyX509TrustManager() };
  507. try {
  508. conn.configure(null, trustAllCerts, null);
  509. conn.setHostnameVerifier(new DummyHostnameVerifier());
  510. } catch (KeyManagementException e) {
  511. throw new IOException(e.getMessage());
  512. } catch (NoSuchAlgorithmException e) {
  513. throw new IOException(e.getMessage());
  514. }
  515. }
  516. final InputStream openInputStream(HttpConnection conn)
  517. throws IOException {
  518. InputStream input = conn.getInputStream();
  519. if (ENCODING_GZIP.equals(conn.getHeaderField(HDR_CONTENT_ENCODING)))
  520. input = new GZIPInputStream(input);
  521. return input;
  522. }
  523. IOException wrongContentType(String expType, String actType) {
  524. final String why = MessageFormat.format(JGitText.get().expectedReceivedContentType, expType, actType);
  525. return new TransportException(uri, why);
  526. }
  527. private boolean isSmartHttp(final HttpConnection c, final String service) {
  528. final String expType = "application/x-" + service + "-advertisement"; //$NON-NLS-1$ //$NON-NLS-2$
  529. final String actType = c.getContentType();
  530. return expType.equals(actType);
  531. }
  532. private void readSmartHeaders(final InputStream in, final String service)
  533. throws IOException {
  534. // A smart reply will have a '#' after the first 4 bytes, but
  535. // a dumb reply cannot contain a '#' until after byte 41. Do a
  536. // quick check to make sure its a smart reply before we parse
  537. // as a pkt-line stream.
  538. //
  539. final byte[] magic = new byte[5];
  540. IO.readFully(in, magic, 0, magic.length);
  541. if (magic[4] != '#') {
  542. throw new TransportException(uri, MessageFormat.format(
  543. JGitText.get().expectedPktLineWithService, RawParseUtils.decode(magic)));
  544. }
  545. final PacketLineIn pckIn = new PacketLineIn(new UnionInputStream(
  546. new ByteArrayInputStream(magic), in));
  547. final String exp = "# service=" + service; //$NON-NLS-1$
  548. final String act = pckIn.readString();
  549. if (!exp.equals(act)) {
  550. throw new TransportException(uri, MessageFormat.format(
  551. JGitText.get().expectedGot, exp, act));
  552. }
  553. while (pckIn.readString() != PacketLineIn.END) {
  554. // for now, ignore the remaining header lines
  555. }
  556. }
  557. class HttpObjectDB extends WalkRemoteObjectDatabase {
  558. private final URL httpObjectsUrl;
  559. HttpObjectDB(final URL b) {
  560. httpObjectsUrl = b;
  561. }
  562. @Override
  563. URIish getURI() {
  564. return new URIish(httpObjectsUrl);
  565. }
  566. @Override
  567. Collection<WalkRemoteObjectDatabase> getAlternates() throws IOException {
  568. try {
  569. return readAlternates(INFO_HTTP_ALTERNATES);
  570. } catch (FileNotFoundException err) {
  571. // Fall through.
  572. }
  573. try {
  574. return readAlternates(INFO_ALTERNATES);
  575. } catch (FileNotFoundException err) {
  576. // Fall through.
  577. }
  578. return null;
  579. }
  580. @Override
  581. WalkRemoteObjectDatabase openAlternate(final String location)
  582. throws IOException {
  583. return new HttpObjectDB(new URL(httpObjectsUrl, location));
  584. }
  585. @Override
  586. Collection<String> getPackNames() throws IOException {
  587. final Collection<String> packs = new ArrayList<String>();
  588. try {
  589. final BufferedReader br = openReader(INFO_PACKS);
  590. try {
  591. for (;;) {
  592. final String s = br.readLine();
  593. if (s == null || s.length() == 0)
  594. break;
  595. if (!s.startsWith("P pack-") || !s.endsWith(".pack")) //$NON-NLS-1$ //$NON-NLS-2$
  596. throw invalidAdvertisement(s);
  597. packs.add(s.substring(2));
  598. }
  599. return packs;
  600. } finally {
  601. br.close();
  602. }
  603. } catch (FileNotFoundException err) {
  604. return packs;
  605. }
  606. }
  607. @Override
  608. FileStream open(final String path) throws IOException {
  609. final URL base = httpObjectsUrl;
  610. final URL u = new URL(base, path);
  611. final HttpConnection c = httpOpen(u);
  612. switch (HttpSupport.response(c)) {
  613. case HttpConnection.HTTP_OK:
  614. final InputStream in = openInputStream(c);
  615. final int len = c.getContentLength();
  616. return new FileStream(in, len);
  617. case HttpConnection.HTTP_NOT_FOUND:
  618. throw new FileNotFoundException(u.toString());
  619. default:
  620. throw new IOException(u.toString() + ": " //$NON-NLS-1$
  621. + HttpSupport.response(c) + " " //$NON-NLS-1$
  622. + c.getResponseMessage());
  623. }
  624. }
  625. Map<String, Ref> readAdvertisedImpl(final BufferedReader br)
  626. throws IOException, PackProtocolException {
  627. final TreeMap<String, Ref> avail = new TreeMap<String, Ref>();
  628. for (;;) {
  629. String line = br.readLine();
  630. if (line == null)
  631. break;
  632. final int tab = line.indexOf('\t');
  633. if (tab < 0)
  634. throw invalidAdvertisement(line);
  635. String name;
  636. final ObjectId id;
  637. name = line.substring(tab + 1);
  638. id = ObjectId.fromString(line.substring(0, tab));
  639. if (name.endsWith("^{}")) { //$NON-NLS-1$
  640. name = name.substring(0, name.length() - 3);
  641. final Ref prior = avail.get(name);
  642. if (prior == null)
  643. throw outOfOrderAdvertisement(name);
  644. if (prior.getPeeledObjectId() != null)
  645. throw duplicateAdvertisement(name + "^{}"); //$NON-NLS-1$
  646. avail.put(name, new ObjectIdRef.PeeledTag(
  647. Ref.Storage.NETWORK, name,
  648. prior.getObjectId(), id));
  649. } else {
  650. Ref prior = avail.put(name, new ObjectIdRef.PeeledNonTag(
  651. Ref.Storage.NETWORK, name, id));
  652. if (prior != null)
  653. throw duplicateAdvertisement(name);
  654. }
  655. }
  656. return avail;
  657. }
  658. private PackProtocolException outOfOrderAdvertisement(final String n) {
  659. return new PackProtocolException(MessageFormat.format(JGitText.get().advertisementOfCameBefore, n, n));
  660. }
  661. private PackProtocolException invalidAdvertisement(final String n) {
  662. return new PackProtocolException(MessageFormat.format(JGitText.get().invalidAdvertisementOf, n));
  663. }
  664. private PackProtocolException duplicateAdvertisement(final String n) {
  665. return new PackProtocolException(MessageFormat.format(JGitText.get().duplicateAdvertisementsOf, n));
  666. }
  667. @Override
  668. void close() {
  669. // We do not maintain persistent connections.
  670. }
  671. }
  672. class SmartHttpFetchConnection extends BasePackFetchConnection {
  673. private MultiRequestService svc;
  674. SmartHttpFetchConnection(final InputStream advertisement)
  675. throws TransportException {
  676. super(TransportHttp.this);
  677. statelessRPC = true;
  678. init(advertisement, DisabledOutputStream.INSTANCE);
  679. outNeedsEnd = false;
  680. readAdvertisedRefs();
  681. }
  682. @Override
  683. protected void doFetch(final ProgressMonitor monitor,
  684. final Collection<Ref> want, final Set<ObjectId> have,
  685. final OutputStream outputStream) throws TransportException {
  686. try {
  687. svc = new MultiRequestService(SVC_UPLOAD_PACK);
  688. init(svc.getInputStream(), svc.getOutputStream());
  689. super.doFetch(monitor, want, have, outputStream);
  690. } finally {
  691. svc = null;
  692. }
  693. }
  694. @Override
  695. protected void onReceivePack() {
  696. svc.finalRequest = true;
  697. }
  698. }
  699. class SmartHttpPushConnection extends BasePackPushConnection {
  700. SmartHttpPushConnection(final InputStream advertisement)
  701. throws TransportException {
  702. super(TransportHttp.this);
  703. statelessRPC = true;
  704. init(advertisement, DisabledOutputStream.INSTANCE);
  705. outNeedsEnd = false;
  706. readAdvertisedRefs();
  707. }
  708. protected void doPush(final ProgressMonitor monitor,
  709. final Map<String, RemoteRefUpdate> refUpdates,
  710. OutputStream outputStream) throws TransportException {
  711. final Service svc = new MultiRequestService(SVC_RECEIVE_PACK);
  712. init(svc.getInputStream(), svc.getOutputStream());
  713. super.doPush(monitor, refUpdates, outputStream);
  714. }
  715. }
  716. /** Basic service for sending and receiving HTTP requests. */
  717. abstract class Service {
  718. protected final String serviceName;
  719. protected final String requestType;
  720. protected final String responseType;
  721. protected HttpConnection conn;
  722. protected HttpOutputStream out;
  723. protected final HttpExecuteStream execute;
  724. final UnionInputStream in;
  725. Service(String serviceName) {
  726. this.serviceName = serviceName;
  727. this.requestType = "application/x-" + serviceName + "-request"; //$NON-NLS-1$ //$NON-NLS-2$
  728. this.responseType = "application/x-" + serviceName + "-result"; //$NON-NLS-1$ //$NON-NLS-2$
  729. this.out = new HttpOutputStream();
  730. this.execute = new HttpExecuteStream();
  731. this.in = new UnionInputStream(execute);
  732. }
  733. void openStream() throws IOException {
  734. conn = httpOpen(METHOD_POST, new URL(baseUrl, serviceName));
  735. conn.setInstanceFollowRedirects(false);
  736. conn.setDoOutput(true);
  737. conn.setRequestProperty(HDR_CONTENT_TYPE, requestType);
  738. conn.setRequestProperty(HDR_ACCEPT, responseType);
  739. }
  740. void sendRequest() throws IOException {
  741. // Try to compress the content, but only if that is smaller.
  742. TemporaryBuffer buf = new TemporaryBuffer.Heap(http.postBuffer);
  743. try {
  744. GZIPOutputStream gzip = new GZIPOutputStream(buf);
  745. out.writeTo(gzip, null);
  746. gzip.close();
  747. if (out.length() < buf.length())
  748. buf = out;
  749. } catch (IOException err) {
  750. // Most likely caused by overflowing the buffer, meaning
  751. // its larger if it were compressed. Don't compress.
  752. buf = out;
  753. }
  754. openStream();
  755. if (buf != out)
  756. conn.setRequestProperty(HDR_CONTENT_ENCODING, ENCODING_GZIP);
  757. conn.setFixedLengthStreamingMode((int) buf.length());
  758. final OutputStream httpOut = conn.getOutputStream();
  759. try {
  760. buf.writeTo(httpOut, null);
  761. } finally {
  762. httpOut.close();
  763. }
  764. }
  765. void openResponse() throws IOException {
  766. final int status = HttpSupport.response(conn);
  767. if (status != HttpConnection.HTTP_OK) {
  768. throw new TransportException(uri, status + " " //$NON-NLS-1$
  769. + conn.getResponseMessage());
  770. }
  771. final String contentType = conn.getContentType();
  772. if (!responseType.equals(contentType)) {
  773. conn.getInputStream().close();
  774. throw wrongContentType(responseType, contentType);
  775. }
  776. }
  777. HttpOutputStream getOutputStream() {
  778. return out;
  779. }
  780. InputStream getInputStream() {
  781. return in;
  782. }
  783. abstract void execute() throws IOException;
  784. class HttpExecuteStream extends InputStream {
  785. public int read() throws IOException {
  786. execute();
  787. return -1;
  788. }
  789. public int read(byte[] b, int off, int len) throws IOException {
  790. execute();
  791. return -1;
  792. }
  793. public long skip(long n) throws IOException {
  794. execute();
  795. return 0;
  796. }
  797. }
  798. class HttpOutputStream extends TemporaryBuffer {
  799. HttpOutputStream() {
  800. super(http.postBuffer);
  801. }
  802. @Override
  803. protected OutputStream overflow() throws IOException {
  804. openStream();
  805. conn.setChunkedStreamingMode(0);
  806. return conn.getOutputStream();
  807. }
  808. }
  809. }
  810. /**
  811. * State required to speak multiple HTTP requests with the remote.
  812. * <p>
  813. * A service wrapper provides a normal looking InputStream and OutputStream
  814. * pair which are connected via HTTP to the named remote service. Writing to
  815. * the OutputStream is buffered until either the buffer overflows, or
  816. * reading from the InputStream occurs. If overflow occurs HTTP/1.1 and its
  817. * chunked transfer encoding is used to stream the request data to the
  818. * remote service. If the entire request fits in the memory buffer, the
  819. * older HTTP/1.0 standard and a fixed content length is used instead.
  820. * <p>
  821. * It is an error to attempt to read without there being outstanding data
  822. * ready for transmission on the OutputStream.
  823. * <p>
  824. * No state is preserved between write-read request pairs. The caller is
  825. * responsible for replaying state vector information as part of the request
  826. * data written to the OutputStream. Any session HTTP cookies may or may not
  827. * be preserved between requests, it is left up to the JVM's implementation
  828. * of the HTTP client.
  829. */
  830. class MultiRequestService extends Service {
  831. boolean finalRequest;
  832. MultiRequestService(final String serviceName) {
  833. super(serviceName);
  834. }
  835. /** Keep opening send-receive pairs to the given URI. */
  836. @Override
  837. void execute() throws IOException {
  838. out.close();
  839. if (conn == null) {
  840. if (out.length() == 0) {
  841. // Request output hasn't started yet, but more data is being
  842. // requested. If there is no request data buffered and the
  843. // final request was already sent, do nothing to ensure the
  844. // caller is shown EOF on the InputStream; otherwise an
  845. // programming error has occurred within this module.
  846. if (finalRequest)
  847. return;
  848. throw new TransportException(uri,
  849. JGitText.get().startingReadStageWithoutWrittenRequestDataPendingIsNotSupported);
  850. }
  851. sendRequest();
  852. }
  853. out.reset();
  854. openResponse();
  855. in.add(openInputStream(conn));
  856. if (!finalRequest)
  857. in.add(execute);
  858. conn = null;
  859. }
  860. }
  861. /** Service for maintaining a single long-poll connection. */
  862. class LongPollService extends Service {
  863. /**
  864. * @param serviceName
  865. */
  866. LongPollService(String serviceName) {
  867. super(serviceName);
  868. }
  869. /** Only open one send-receive request. */
  870. @Override
  871. void execute() throws IOException {
  872. out.close();
  873. if (conn == null)
  874. sendRequest();
  875. openResponse();
  876. in.add(openInputStream(conn));
  877. }
  878. }
  879. private static class DummyX509TrustManager implements X509TrustManager {
  880. public X509Certificate[] getAcceptedIssuers() {
  881. return null;
  882. }
  883. public void checkClientTrusted(X509Certificate[] certs, String authType) {
  884. // no check
  885. }
  886. public void checkServerTrusted(X509Certificate[] certs, String authType) {
  887. // no check
  888. }
  889. }
  890. private static class DummyHostnameVerifier implements HostnameVerifier {
  891. public boolean verify(String hostname, SSLSession session) {
  892. // always accept
  893. return true;
  894. }
  895. }
  896. }