Line data Source code
1 : // Protocol Buffers - Google's data interchange format
2 : // Copyright 2008 Google Inc. All rights reserved.
3 : // https://developers.google.com/protocol-buffers/
4 : //
5 : // Redistribution and use in source and binary forms, with or without
6 : // modification, are permitted provided that the following conditions are
7 : // met:
8 : //
9 : // * Redistributions of source code must retain the above copyright
10 : // notice, this list of conditions and the following disclaimer.
11 : // * Redistributions in binary form must reproduce the above
12 : // copyright notice, this list of conditions and the following disclaimer
13 : // in the documentation and/or other materials provided with the
14 : // distribution.
15 : // * Neither the name of Google Inc. nor the names of its
16 : // contributors may be used to endorse or promote products derived from
17 : // this software without specific prior written permission.
18 : //
19 : // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 : // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 : // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 : // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 : // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 : // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 : // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 : // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 : // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 : // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 : // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30 :
31 : // Author: kenton@google.com (Kenton Varda)
32 : // Based on original Protocol Buffers design by
33 : // Sanjay Ghemawat, Jeff Dean, and others.
34 :
35 : #ifdef _MSC_VER
36 : #include <io.h>
37 : #else
38 : #include <unistd.h>
39 : #endif
40 : #include <sys/types.h>
41 : #include <sys/stat.h>
42 : #include <fcntl.h>
43 : #include <errno.h>
44 :
45 : #include <algorithm>
46 : #include <memory>
47 : #ifndef _SHARED_PTR_H
48 : #include <google/protobuf/stubs/shared_ptr.h>
49 : #endif
50 :
51 : #include <google/protobuf/compiler/importer.h>
52 :
53 : #include <google/protobuf/compiler/parser.h>
54 : #include <google/protobuf/io/tokenizer.h>
55 : #include <google/protobuf/io/zero_copy_stream_impl.h>
56 : #include <google/protobuf/stubs/strutil.h>
57 :
58 : namespace google {
59 : namespace protobuf {
60 : namespace compiler {
61 :
62 : #ifdef _WIN32
63 : #ifndef F_OK
64 : #define F_OK 00 // not defined by MSVC for whatever reason
65 : #endif
66 : #include <ctype.h>
67 : #endif
68 :
69 : // Returns true if the text looks like a Windows-style absolute path, starting
70 : // with a drive letter. Example: "C:\foo". TODO(kenton): Share this with
71 : // copy in command_line_interface.cc?
72 : static bool IsWindowsAbsolutePath(const string& text) {
73 : #if defined(_WIN32) || defined(__CYGWIN__)
74 : return text.size() >= 3 && text[1] == ':' &&
75 : isalpha(text[0]) &&
76 : (text[2] == '/' || text[2] == '\\') &&
77 : text.find_last_of(':') == 1;
78 : #else
79 : return false;
80 : #endif
81 : }
82 :
83 17 : MultiFileErrorCollector::~MultiFileErrorCollector() {}
84 :
85 : // This class serves two purposes:
86 : // - It implements the ErrorCollector interface (used by Tokenizer and Parser)
87 : // in terms of MultiFileErrorCollector, using a particular filename.
88 : // - It lets us check if any errors have occurred.
89 : class SourceTreeDescriptorDatabase::SingleFileErrorCollector
90 : : public io::ErrorCollector {
91 : public:
92 78 : SingleFileErrorCollector(const string& filename,
93 : MultiFileErrorCollector* multi_file_error_collector)
94 : : filename_(filename),
95 : multi_file_error_collector_(multi_file_error_collector),
96 156 : had_errors_(false) {}
97 156 : ~SingleFileErrorCollector() {}
98 :
99 : bool had_errors() { return had_errors_; }
100 :
101 : // implements ErrorCollector ---------------------------------------
102 0 : void AddError(int line, int column, const string& message) {
103 0 : if (multi_file_error_collector_ != NULL) {
104 0 : multi_file_error_collector_->AddError(filename_, line, column, message);
105 : }
106 0 : had_errors_ = true;
107 0 : }
108 :
109 : private:
110 : string filename_;
111 : MultiFileErrorCollector* multi_file_error_collector_;
112 : bool had_errors_;
113 : };
114 :
115 : // ===================================================================
116 :
117 17 : SourceTreeDescriptorDatabase::SourceTreeDescriptorDatabase(
118 : SourceTree* source_tree)
119 : : source_tree_(source_tree),
120 : error_collector_(NULL),
121 : using_validation_error_collector_(false),
122 34 : validation_error_collector_(this) {}
123 :
124 17 : SourceTreeDescriptorDatabase::~SourceTreeDescriptorDatabase() {}
125 :
126 78 : bool SourceTreeDescriptorDatabase::FindFileByName(
127 : const string& filename, FileDescriptorProto* output) {
128 78 : google::protobuf::scoped_ptr<io::ZeroCopyInputStream> input(source_tree_->Open(filename));
129 78 : if (input == NULL) {
130 0 : if (error_collector_ != NULL) {
131 : error_collector_->AddError(filename, -1, 0,
132 0 : source_tree_->GetLastErrorMessage());
133 : }
134 : return false;
135 : }
136 :
137 : // Set up the tokenizer and parser.
138 156 : SingleFileErrorCollector file_error_collector(filename, error_collector_);
139 156 : io::Tokenizer tokenizer(input.get(), &file_error_collector);
140 :
141 156 : Parser parser;
142 78 : if (error_collector_ != NULL) {
143 78 : parser.RecordErrorsTo(&file_error_collector);
144 : }
145 78 : if (using_validation_error_collector_) {
146 78 : parser.RecordSourceLocationsTo(&source_locations_);
147 : }
148 :
149 : // Parse it.
150 : output->set_name(filename);
151 156 : return parser.Parse(&tokenizer, output) &&
152 78 : !file_error_collector.had_errors();
153 : }
154 :
155 58 : bool SourceTreeDescriptorDatabase::FindFileContainingSymbol(
156 : const string& symbol_name, FileDescriptorProto* output) {
157 58 : return false;
158 : }
159 :
160 0 : bool SourceTreeDescriptorDatabase::FindFileContainingExtension(
161 : const string& containing_type, int field_number,
162 : FileDescriptorProto* output) {
163 0 : return false;
164 : }
165 :
166 : // -------------------------------------------------------------------
167 :
168 17 : SourceTreeDescriptorDatabase::ValidationErrorCollector::
169 : ValidationErrorCollector(SourceTreeDescriptorDatabase* owner)
170 34 : : owner_(owner) {}
171 :
172 17 : SourceTreeDescriptorDatabase::ValidationErrorCollector::
173 17 : ~ValidationErrorCollector() {}
174 :
175 0 : void SourceTreeDescriptorDatabase::ValidationErrorCollector::AddError(
176 : const string& filename,
177 : const string& element_name,
178 : const Message* descriptor,
179 : ErrorLocation location,
180 : const string& message) {
181 0 : if (owner_->error_collector_ == NULL) return;
182 :
183 : int line, column;
184 0 : owner_->source_locations_.Find(descriptor, location, &line, &column);
185 0 : owner_->error_collector_->AddError(filename, line, column, message);
186 : }
187 :
188 : // ===================================================================
189 :
190 17 : Importer::Importer(SourceTree* source_tree,
191 : MultiFileErrorCollector* error_collector)
192 : : database_(source_tree),
193 34 : pool_(&database_, database_.GetValidationErrorCollector()) {
194 17 : pool_.EnforceWeakDependencies(true);
195 17 : database_.RecordErrorsTo(error_collector);
196 17 : }
197 :
198 17 : Importer::~Importer() {}
199 :
200 57 : const FileDescriptor* Importer::Import(const string& filename) {
201 57 : return pool_.FindFileByName(filename);
202 : }
203 :
204 57 : void Importer::AddUnusedImportTrackFile(const string& file_name) {
205 57 : pool_.AddUnusedImportTrackFile(file_name);
206 57 : }
207 :
208 57 : void Importer::ClearUnusedImportTrackFiles() {
209 57 : pool_.ClearUnusedImportTrackFiles();
210 57 : }
211 :
212 : // ===================================================================
213 :
214 17 : SourceTree::~SourceTree() {}
215 :
216 0 : string SourceTree::GetLastErrorMessage() {
217 0 : return "File not found.";
218 : }
219 :
220 68 : DiskSourceTree::DiskSourceTree() {}
221 :
222 34 : DiskSourceTree::~DiskSourceTree() {}
223 :
224 : static inline char LastChar(const string& str) {
225 306 : return str[str.size() - 1];
226 : }
227 :
228 : // Given a path, returns an equivalent path with these changes:
229 : // - On Windows, any backslashes are replaced with forward slashes.
230 : // - Any instances of the directory "." are removed.
231 : // - Any consecutive '/'s are collapsed into a single slash.
232 : // Note that the resulting string may be empty.
233 : //
234 : // TODO(kenton): It would be nice to handle "..", e.g. so that we can figure
235 : // out that "foo/bar.proto" is inside "baz/../foo". However, if baz is a
236 : // symlink or doesn't exist, then things get complicated, and we can't
237 : // actually determine this without investigating the filesystem, probably
238 : // in non-portable ways. So, we punt.
239 : //
240 : // TODO(kenton): It would be nice to use realpath() here except that it
241 : // resolves symbolic links. This could cause problems if people place
242 : // symbolic links in their source tree. For example, if you executed:
243 : // protoc --proto_path=foo foo/bar/baz.proto
244 : // then if foo/bar is a symbolic link, foo/bar/baz.proto will canonicalize
245 : // to a path which does not appear to be under foo, and thus the compiler
246 : // will complain that baz.proto is not inside the --proto_path.
247 153 : static string CanonicalizePath(string path) {
248 : #ifdef _WIN32
249 : // The Win32 API accepts forward slashes as a path delimiter even though
250 : // backslashes are standard. Let's avoid confusion and use only forward
251 : // slashes.
252 : if (HasPrefixString(path, "\\\\")) {
253 : // Avoid converting two leading backslashes.
254 : path = "\\\\" + StringReplace(path.substr(2), "\\", "/", true);
255 : } else {
256 : path = StringReplace(path, "\\", "/", true);
257 : }
258 : #endif
259 :
260 : vector<string> canonical_parts;
261 : vector<string> parts = Split(
262 306 : path, "/", true); // Note: Removes empty parts.
263 1191 : for (int i = 0; i < parts.size(); i++) {
264 1557 : if (parts[i] == ".") {
265 : // Ignore.
266 : } else {
267 1004 : canonical_parts.push_back(parts[i]);
268 : }
269 : }
270 153 : string result = Join(canonical_parts, "/");
271 306 : if (!path.empty() && path[0] == '/') {
272 : // Restore leading slash.
273 2 : result = '/' + result;
274 : }
275 306 : if (!path.empty() && LastChar(path) == '/' &&
276 153 : !result.empty() && LastChar(result) != '/') {
277 : // Restore trailing slash.
278 : result += '/';
279 : }
280 153 : return result;
281 : }
282 :
283 213 : static inline bool ContainsParentReference(const string& path) {
284 213 : return path == ".." ||
285 1065 : HasPrefixString(path, "../") ||
286 1065 : HasSuffixString(path, "/..") ||
287 426 : path.find("/../") != string::npos;
288 : }
289 :
290 : // Maps a file from an old location to a new one. Typically, old_prefix is
291 : // a virtual path and new_prefix is its corresponding disk path. Returns
292 : // false if the filename did not start with old_prefix, otherwise replaces
293 : // old_prefix with new_prefix and stores the result in *result. Examples:
294 : // string result;
295 : // assert(ApplyMapping("foo/bar", "", "baz", &result));
296 : // assert(result == "baz/foo/bar");
297 : //
298 : // assert(ApplyMapping("foo/bar", "foo", "baz", &result));
299 : // assert(result == "baz/bar");
300 : //
301 : // assert(ApplyMapping("foo", "foo", "bar", &result));
302 : // assert(result == "bar");
303 : //
304 : // assert(!ApplyMapping("foo/bar", "baz", "qux", &result));
305 : // assert(!ApplyMapping("foo/bar", "baz", "qux", &result));
306 : // assert(!ApplyMapping("foobar", "foo", "baz", &result));
307 135 : static bool ApplyMapping(const string& filename,
308 : const string& old_prefix,
309 : const string& new_prefix,
310 : string* result) {
311 135 : if (old_prefix.empty()) {
312 : // old_prefix matches any relative path.
313 135 : if (ContainsParentReference(filename)) {
314 : // We do not allow the file name to use "..".
315 : return false;
316 : }
317 270 : if (HasPrefixString(filename, "/") ||
318 135 : IsWindowsAbsolutePath(filename)) {
319 : // This is an absolute path, so it isn't matched by the empty string.
320 : return false;
321 : }
322 135 : result->assign(new_prefix);
323 135 : if (!result->empty()) result->push_back('/');
324 135 : result->append(filename);
325 135 : return true;
326 0 : } else if (HasPrefixString(filename, old_prefix)) {
327 : // old_prefix is a prefix of the filename. Is it the whole filename?
328 0 : if (filename.size() == old_prefix.size()) {
329 : // Yep, it's an exact match.
330 : *result = new_prefix;
331 0 : return true;
332 : } else {
333 : // Not an exact match. Is the next character a '/'? Otherwise,
334 : // this isn't actually a match at all. E.g. the prefix "foo/bar"
335 : // does not match the filename "foo/barbaz".
336 0 : int after_prefix_start = -1;
337 0 : if (filename[old_prefix.size()] == '/') {
338 0 : after_prefix_start = old_prefix.size() + 1;
339 0 : } else if (filename[old_prefix.size() - 1] == '/') {
340 : // old_prefix is never empty, and canonicalized paths never have
341 : // consecutive '/' characters.
342 0 : after_prefix_start = old_prefix.size();
343 : }
344 0 : if (after_prefix_start != -1) {
345 : // Yep. So the prefixes are directories and the filename is a file
346 : // inside them.
347 0 : string after_prefix = filename.substr(after_prefix_start);
348 0 : if (ContainsParentReference(after_prefix)) {
349 : // We do not allow the file name to use "..".
350 : return false;
351 : }
352 0 : result->assign(new_prefix);
353 0 : if (!result->empty()) result->push_back('/');
354 0 : result->append(after_prefix);
355 : return true;
356 : }
357 : }
358 : }
359 :
360 : return false;
361 : }
362 :
363 18 : void DiskSourceTree::MapPath(const string& virtual_path,
364 : const string& disk_path) {
365 54 : mappings_.push_back(Mapping(virtual_path, CanonicalizePath(disk_path)));
366 18 : }
367 :
368 : DiskSourceTree::DiskFileToVirtualFileResult
369 57 : DiskSourceTree::DiskFileToVirtualFile(
370 : const string& disk_file,
371 : string* virtual_file,
372 : string* shadowing_disk_file) {
373 57 : int mapping_index = -1;
374 114 : string canonical_disk_file = CanonicalizePath(disk_file);
375 :
376 228 : for (int i = 0; i < mappings_.size(); i++) {
377 : // Apply the mapping in reverse.
378 114 : if (ApplyMapping(canonical_disk_file, mappings_[i].disk_path,
379 285 : mappings_[i].virtual_path, virtual_file)) {
380 : // Success.
381 : mapping_index = i;
382 : break;
383 : }
384 : }
385 :
386 57 : if (mapping_index == -1) {
387 : return NO_MAPPING;
388 : }
389 :
390 : // Iterate through all mappings with higher precedence and verify that none
391 : // of them map this file to some other existing file.
392 0 : for (int i = 0; i < mapping_index; i++) {
393 0 : if (ApplyMapping(*virtual_file, mappings_[i].virtual_path,
394 0 : mappings_[i].disk_path, shadowing_disk_file)) {
395 0 : if (access(shadowing_disk_file->c_str(), F_OK) >= 0) {
396 : // File exists.
397 : return SHADOWED;
398 : }
399 : }
400 : }
401 : shadowing_disk_file->clear();
402 :
403 : // Verify that we can open the file. Note that this also has the side-effect
404 : // of verifying that we are not canonicalizing away any non-existent
405 : // directories.
406 57 : google::protobuf::scoped_ptr<io::ZeroCopyInputStream> stream(OpenDiskFile(disk_file));
407 57 : if (stream == NULL) {
408 : return CANNOT_OPEN;
409 : }
410 :
411 57 : return SUCCESS;
412 : }
413 :
414 0 : bool DiskSourceTree::VirtualFileToDiskFile(const string& virtual_file,
415 : string* disk_file) {
416 : google::protobuf::scoped_ptr<io::ZeroCopyInputStream> stream(
417 0 : OpenVirtualFile(virtual_file, disk_file));
418 0 : return stream != NULL;
419 : }
420 :
421 78 : io::ZeroCopyInputStream* DiskSourceTree::Open(const string& filename) {
422 78 : return OpenVirtualFile(filename, NULL);
423 : }
424 :
425 0 : string DiskSourceTree::GetLastErrorMessage() {
426 0 : return last_error_message_;
427 : }
428 :
429 78 : io::ZeroCopyInputStream* DiskSourceTree::OpenVirtualFile(
430 : const string& virtual_file,
431 : string* disk_file) {
432 390 : if (virtual_file != CanonicalizePath(virtual_file) ||
433 78 : ContainsParentReference(virtual_file)) {
434 : // We do not allow importing of paths containing things like ".." or
435 : // consecutive slashes since the compiler expects files to be uniquely
436 : // identified by file name.
437 0 : last_error_message_ = "Backslashes, consecutive slashes, \".\", or \"..\" "
438 : "are not allowed in the virtual path";
439 0 : return NULL;
440 : }
441 :
442 312 : for (int i = 0; i < mappings_.size(); i++) {
443 : string temp_disk_file;
444 156 : if (ApplyMapping(virtual_file, mappings_[i].virtual_path,
445 390 : mappings_[i].disk_path, &temp_disk_file)) {
446 78 : io::ZeroCopyInputStream* stream = OpenDiskFile(temp_disk_file);
447 78 : if (stream != NULL) {
448 78 : if (disk_file != NULL) {
449 : *disk_file = temp_disk_file;
450 : }
451 78 : return stream;
452 : }
453 :
454 0 : if (errno == EACCES) {
455 : // The file exists but is not readable.
456 0 : last_error_message_ = "Read access is denied for file: " +
457 : temp_disk_file;
458 0 : return NULL;
459 : }
460 : }
461 : }
462 0 : last_error_message_ = "File not found.";
463 0 : return NULL;
464 : }
465 :
466 135 : io::ZeroCopyInputStream* DiskSourceTree::OpenDiskFile(
467 : const string& filename) {
468 : int file_descriptor;
469 135 : do {
470 270 : file_descriptor = open(filename.c_str(), O_RDONLY);
471 0 : } while (file_descriptor < 0 && errno == EINTR);
472 135 : if (file_descriptor >= 0) {
473 135 : io::FileInputStream* result = new io::FileInputStream(file_descriptor);
474 : result->SetCloseOnDelete(true);
475 135 : return result;
476 : } else {
477 : return NULL;
478 : }
479 : }
480 :
481 : } // namespace compiler
482 : } // namespace protobuf
483 : } // namespace google
|