Purify CanonPath

src/libutil/canon-path.cc

@@ -1,16 +1,25 @@
 #include "canon-path.hh"
-#include "file-system.hh"
+#include "util.hh"
+#include "file-path-impl.hh"
 
 namespace nix {
 
 CanonPath CanonPath::root = CanonPath("/");
 
+static std::string absPathPure(std::string_view path)
+{
+    return canonPathInner(path, [](auto &, auto &){});
+}
+
 CanonPath::CanonPath(std::string_view raw)
-    : path(absPath(raw, "/"))
+    : path(absPathPure(concatStrings("/", raw)))
 { }
 
 CanonPath::CanonPath(std::string_view raw, const CanonPath & root)
-    : path(absPath(raw, root.abs()))
+    : path(absPathPure(
+        raw.size() > 0 && raw[0] == '/'
+            ? raw
+            : concatStrings(root.abs(), "/", raw)))
 { }
 
 CanonPath::CanonPath(const std::vector<std::string> & elems)

src/libutil/canon-path.hh

@@ -21,9 +21,21 @@ namespace nix {
  *
  * - There are no components equal to '.' or '..'.
  *
- * Note that the path does not need to correspond to an actually
- * existing path, and there is no guarantee that symlinks are
- * resolved.
+ * `CanonPath` are "virtual" Nix paths for abstract file system objects;
+ * they are always Unix-style paths, regardless of what OS Nix is
+ * running on. The `/` root doesn't denote the ambient host file system
+ * root, but some virtual FS root.
+ *
+ * @note It might be useful to compare `openat(some_fd, "foo/bar")` on
+ * Unix. `"foo/bar"` is a relative path because an absolute path would
+ * "override" the `some_fd` directory file descriptor and escape to the
+ * "system root". Conversely, Nix's abstract file operations *never* escape the
+ * designated virtual file system (i.e. `SourceAccessor` or
+ * `ParseSink`), so `CanonPath` does not need an absolute/relative
+ * distinction.
+ *
+ * @note The path does not need to correspond to an actually existing
+ * path, and the path may or may not have unresolved symlinks.
  */
 class CanonPath
 {

src/libutil/file-path-impl.hh

@@ -0,0 +1,81 @@
+#pragma once
+/**
+ * @file
+ *
+ * Pure (no IO) infrastructure just for defining other path types;
+ * should not be used directly outside of utilities.
+ */
+#include <string>
+#include <string_view>
+
+namespace nix {
+
+/**
+ * Core pure path canonicalization algorithm.
+ *
+ * @param hookComponent
+ *   A callback which is passed two arguments,
+ *   references to
+ *
+ *   1. the result so far
+ *
+ *   2. the remaining path to resolve
+ *
+ *   This is a chance to modify those two paths in arbitrary way, e.g. if
+ *   "result" points to a symlink.
+ */
+typename std::string canonPathInner(
+    std::string_view remaining,
+    auto && hookComponent)
+{
+    assert(remaining != "");
+
+    std::string result;
+    result.reserve(256);
+
+    while (true) {
+
+        /* Skip slashes. */
+        while (!remaining.empty() && remaining[0] == '/')
+            remaining.remove_prefix(1);
+
+        if (remaining.empty()) break;
+
+        auto nextComp = ({
+            auto nextPathSep = remaining.find('/');
+            nextPathSep == remaining.npos ? remaining : remaining.substr(0, nextPathSep);
+        });
+
+        /* Ignore `.'. */
+        if (nextComp == ".")
+            remaining.remove_prefix(1);
+
+        /* If `..', delete the last component. */
+        else if (nextComp == "..")
+        {
+            if (!result.empty()) result.erase(result.rfind('/'));
+            remaining.remove_prefix(2);
+        }
+
+        /* Normal component; copy it. */
+        else {
+            result += '/';
+            if (const auto slash = remaining.find('/'); slash == result.npos) {
+                result += remaining;
+                remaining = {};
+            } else {
+                result += remaining.substr(0, slash);
+                remaining = remaining.substr(slash);
+            }
+
+            hookComponent(result, remaining);
+        }
+    }
+
+    if (result.empty())
+        result = "/";
+
+    return result;
+}
+
+}

src/libutil/file-system.cc

@@ -1,5 +1,6 @@
 #include "environment-variables.hh"
 #include "file-system.hh"
+#include "file-path-impl.hh"
 #include "signals.hh"
 #include "finally.hh"
 #include "serialise.hh"
@@ -21,11 +22,18 @@ namespace fs = std::filesystem;
 
 namespace nix {
 
+/** Treat the string as possibly an absolute path, by inspecting the start of it. Return whether it was probably intended to be absolute. */
+static bool isAbsolute(PathView path)
+{
+    return !path.empty() && path[0] == '/';
+}
+
+
 Path absPath(PathView path, std::optional<PathView> dir, bool resolveSymlinks)
 {
     std::string scratch;
 
-    if (path.empty() || path[0] != '/') {
+    if (!isAbsolute(path)) {
         // In this case we need to call `canonPath` on a newly-created
         // string. We set `scratch` to that string first, and then set
         // `path` to `scratch`. This ensures the newly-created string
@@ -58,69 +66,39 @@ Path canonPath(PathView path, bool resolveSymlinks)
 {
     assert(path != "");
 
-    std::string s;
-    s.reserve(256);
-
-    if (path[0] != '/')
+    if (!isAbsolute(path))
         throw Error("not an absolute path: '%1%'", path);
 
+    /* This just exists because we cannot set the target of `remaining`
+       (the callback parameter) directly to a newly-constructed string,
+       since it is `std::string_view`. */
     std::string temp;
 
     /* Count the number of times we follow a symlink and stop at some
        arbitrary (but high) limit to prevent infinite loops. */
     unsigned int followCount = 0, maxFollow = 1024;
 
-    while (1) {
-
-        /* Skip slashes. */
-        while (!path.empty() && path[0] == '/') path.remove_prefix(1);
-        if (path.empty()) break;
-
-        /* Ignore `.'. */
-        if (path == "." || path.substr(0, 2) == "./")
-            path.remove_prefix(1);
-
-        /* If `..', delete the last component. */
-        else if (path == ".." || path.substr(0, 3) == "../")
-        {
-            if (!s.empty()) s.erase(s.rfind('/'));
-            path.remove_prefix(2);
-        }
-
-        /* Normal component; copy it. */
-        else {
-            s += '/';
-            if (const auto slash = path.find('/'); slash == path.npos) {
-                s += path;
-                path = {};
-            } else {
-                s += path.substr(0, slash);
-                path = path.substr(slash);
-            }
-
-            /* If s points to a symlink, resolve it and continue from there */
-            if (resolveSymlinks && isLink(s)) {
+    return canonPathInner(
+        path,
+        [&followCount, &temp, maxFollow, resolveSymlinks]
+        (std::string & result, std::string_view & remaining) {
+            if (resolveSymlinks && isLink(result)) {
                 if (++followCount >= maxFollow)
-                    throw Error("infinite symlink recursion in path '%1%'", path);
-                temp = concatStrings(readLink(s), path);
-                path = temp;
-                if (!temp.empty() && temp[0] == '/') {
-                    s.clear();  /* restart for symlinks pointing to absolute path */
+                    throw Error("infinite symlink recursion in path '%0%'", remaining);
+                remaining = (temp = concatStrings(readLink(result), remaining));
+                if (isAbsolute(remaining)) {
+                    /* restart for symlinks pointing to absolute path */
+                    result.clear();
                 } else {
-                    s = dirOf(s);
-                    if (s == "/") {  // we don’t want trailing slashes here, which dirOf only produces if s = /
-                        s.clear();
+                    result = dirOf(result);
+                    if (result == "/") {
+                        /* we don’t want trailing slashes here, which `dirOf`
+                           only produces if `result = /` */
+                        result.clear();
                     }
                 }
             }
-        }
-    }
-
-    if (s.empty()) {
-        s = "/";
-    }
-
-    return s;
+        });
 }
 
 

tests/unit/libutil/canon-path.cc

@@ -41,6 +41,24 @@ namespace nix {
         }
     }
 
+    TEST(CanonPath, from_existing) {
+        CanonPath p0("foo//bar/");
+        {
+            CanonPath p("/baz//quux/", p0);
+            ASSERT_EQ(p.abs(), "/baz/quux");
+            ASSERT_EQ(p.rel(), "baz/quux");
+            ASSERT_EQ(*p.baseName(), "quux");
+            ASSERT_EQ(*p.dirOf(), "/baz");
+        }
+        {
+            CanonPath p("baz//quux/", p0);
+            ASSERT_EQ(p.abs(), "/foo/bar/baz/quux");
+            ASSERT_EQ(p.rel(), "foo/bar/baz/quux");
+            ASSERT_EQ(*p.baseName(), "quux");
+            ASSERT_EQ(*p.dirOf(), "/foo/bar/baz");
+        }
+    }
+
     TEST(CanonPath, pop) {
         CanonPath p("foo/bar/x");
         ASSERT_EQ(p.abs(), "/foo/bar/x");