From 8272f98ee26144fc8108ca35bc49b7e25868535f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fran=C3=A7ois=20Degros?= <fdegros@chromium.org>
Date: Thu, 17 Sep 2020 18:06:27 +1000
Subject: [PATCH] Deduplicate file names

---
 lib/fileNode.cpp    | 27 ++++++--------------------
 lib/fileNode.h      |  9 ++++-----
 lib/fuse-zip.cpp    |  4 ++--
 lib/fuseZipData.cpp | 47 ++++++++++++++++++++++++++++-----------------
 lib/fuseZipData.h   |  6 +++++-
 5 files changed, 46 insertions(+), 47 deletions(-)

diff --git a/lib/fileNode.cpp b/lib/fileNode.cpp
index 62723d6..8a5b16a 100644
--- a/lib/fileNode.cpp
+++ b/lib/fileNode.cpp
@@ -89,7 +89,6 @@ FileNode *FileNode::createDir(struct zip *zip, const char *fname,
 FileNode *FileNode::createRootNode(struct zip *zip) {
     auto data = DataNode::createNew(S_IFDIR | 0755, 0, 0, 0);
     FileNode *n = new FileNode(zip, "", ROOT_NODE_INDEX, std::move(data));
-    n->name = n->full_name.c_str();
 
     int len = 0;
     n->m_comment = zip_get_archive_comment(zip, &len, ZIP_FL_ENC_RAW);
@@ -137,26 +136,12 @@ FileNode::~FileNode() {
 void FileNode::parse_name() {
     assert(!full_name.empty());
 
-    const char *lsl = full_name.c_str();
-    while (*lsl++) {}
-    lsl--;
-    while (lsl > full_name.c_str() && *lsl != '/') {
-        lsl--;
-    }
     // If the last symbol in file name is '/' then it is a directory
-    if (*lsl == '/' && *(lsl+1) == '\0') {
-        // It will produce two \0s at the end of file name. I think that
-        // it is not a problem
-        full_name[full_name.size() - 1] = 0;
-        while (lsl > full_name.c_str() && *lsl != '/') {
-            lsl--;
-        }
-    }
-    // Setting short name of node
-    if (*lsl == '/') {
-        lsl++;
-    }
-    this->name = lsl;
+    if (full_name.back() == '/')
+        full_name.pop_back();
+
+    name_start = full_name.find_last_of('/') + 1;
+    assert(name_start <= full_name.size());
 }
 
 void FileNode::readComment() {
@@ -372,7 +357,7 @@ int FileNode::updateExternalAttributes() const {
         // FILE_ATTRIBUTE_DIRECTORY
         mode |= 0x10;
     }
-    if (name[0] == '.') {
+    if (full_name[name_start] == '.') {
         // FILE_ATTRIBUTE_HIDDEN
         mode |= 2;
     }
diff --git a/lib/fileNode.h b/lib/fileNode.h
index ab07cd9..e20cd0d 100644
--- a/lib/fileNode.h
+++ b/lib/fileNode.h
@@ -191,10 +191,9 @@ public:
     /**
      * Get parent name
      */
-    //TODO: rewrite without memory allocation
-    inline std::string getParentName () const {
-        if (name > full_name.c_str()) {
-            return std::string (full_name, 0, static_cast<size_t>(name - full_name.c_str() - 1));
+    std::string getParentName() const {
+        if (name_start > 0) {
+            return std::string(full_name, 0, name_start - 1);
         } else {
             return "";
         }
@@ -230,7 +229,7 @@ public:
     const char *getComment() const { return m_comment; }
     uint16_t getCommentLength() const { return m_commentLen; }
 
-    const char *name;
+    std::string::size_type name_start = 0;
     std::string full_name;
     nodelist_t childs;
     FileNode *parent;
diff --git a/lib/fuse-zip.cpp b/lib/fuse-zip.cpp
index 7c33999..5310925 100644
--- a/lib/fuse-zip.cpp
+++ b/lib/fuse-zip.cpp
@@ -167,8 +167,8 @@ int fusezip_readdir(const char *path, void *buf, fuse_fill_dir_t filler, off_t o
     }
     filler(buf, ".", NULL, 0);
     filler(buf, "..", NULL, 0);
-    for (nodelist_t::const_iterator i = node->childs.begin(); i != node->childs.end(); ++i) {
-        filler(buf, (*i)->name, NULL, 0);
+    for (const FileNode *p : node->childs) {
+        filler(buf, p->full_name.c_str() + p->name_start, NULL, 0);
     }
 
     return 0;
diff --git a/lib/fuseZipData.cpp b/lib/fuseZipData.cpp
index 6d5d498..c511e1d 100644
--- a/lib/fuseZipData.cpp
+++ b/lib/fuseZipData.cpp
@@ -298,22 +298,40 @@ mode_t FuseZipData::getEntryAttributes(zip_uint64_t id, const char *name, bool &
     return mode;
 }
 
+void FuseZipData::renameAndAttachNode(FileNode *node) {
+    std::string &f = node->full_name;
+    if (files.insert({f.c_str(), node}).second)
+        return;
+
+    // Duplicate file name
+    const std::string s = std::move(f);
+
+    // Find extension start
+    std::string::size_type e = s.find_last_of('.');
+    if (e <= node->name_start || e >= s.size() - 1)
+        e = s.size();
+
+    // Add a number before the extension
+    int i = 0;
+    do {
+        f.clear();
+        f.append(s, 0, e);
+        f.append(" (");
+        f.append(std::to_string(++i));
+        f.append(")");
+        f.append(s, e);
+    } while (!files.insert({f.c_str(), node}).second);
+}
+
 void FuseZipData::attachNode(zip_int64_t id, const char *name, mode_t mode, bool readonly,
             bool needPrefix, filemap_t &origNames)
 {
     std::string converted;
     convertFileName(name, readonly, needPrefix, converted);
     const char *cname = converted.c_str();
-    if (files.find(cname) != files.end()) {
-        throw ZipError(std::string("Duplicated file name: ") + cname,
-                       ZIP_ER_EXISTS);
-    }
     FileNode *node = FileNode::createNodeForZipEntry(m_zip, cname, id, mode);
-    if (node == NULL) {
-        throw std::bad_alloc();
-    }
-    files[node->full_name.c_str()] = node;
-    origNames[name] = node;
+    renameAndAttachNode(node);
+    origNames.insert({name, node});
 }
 
 bool FuseZipData::attachHardlink(zip_int64_t sid, const char *name, mode_t mode, bool readonly,
@@ -378,16 +396,9 @@ bool FuseZipData::attachHardlink(zip_int64_t sid, const char *name, mode_t mode,
     std::string converted;
     convertFileName(name, readonly, needPrefix, converted);
     const char *cname = converted.c_str();
-    if (files.find(cname) != files.end()) {
-        throw ZipError(std::string("Duplicated file name: ") + cname,
-                       ZIP_ER_EXISTS);
-    }
     FileNode *node = FileNode::createHardlink(m_zip, cname, sid, it->second);
-    if (node == NULL) {
-        throw std::bad_alloc();
-    }
-    files[node->full_name.c_str()] = node;
-    origNames[name] = node;
+    renameAndAttachNode(node);
+    origNames.insert({name, node});
 
     return true;
 }
diff --git a/lib/fuseZipData.h b/lib/fuseZipData.h
index e25c060..92d4191 100644
--- a/lib/fuseZipData.h
+++ b/lib/fuseZipData.h
@@ -66,6 +66,11 @@ private:
      */
     mode_t getEntryAttributes(zip_uint64_t id, const char *name, bool &isHardlink);
 
+    /**
+     * Attach a node, renaming it if necessary to prevent name collisions.
+     */
+    void renameAndAttachNode(FileNode *node);
+
     /**
      * create and attach file node
      */
@@ -143,4 +148,3 @@ public:
 };
 
 #endif
-
-- 
2.29.1.341.ge80a0c044ae-goog