usm/src/lib/Ignore/UsmIgnore.vala

using Invercargill.DataStructures;

namespace Usm {

    /**
     * Matcher for a package root's `.usmignore` file, consulted by
     * `usm manifest package` to prune files and directories from the
     * produced archive.
     *
     * Semantics (there is no negation: `!` is a literal pattern character,
     * it cannot re-include a previously ignored path):
     *
     * - Blank lines and lines starting with `#` are skipped.
     * - A pattern containing `/` matches the FULL path relative to the
     *   package root; a pattern without `/` matches any path suffix
     *   (equivalently, the basename at any depth).
     * - A trailing `/` marks a directory-only pattern: it matches the
     *   directory itself and everything beneath it.
     * - `*` and `?` are wildcards and never match `/`.
     * - The root `.usmignore` and `MANIFEST.usm` files can never be ignored.
     * - The `.git` directory is always ignored, with or without a
     *   `.usmignore` file; when no `.usmignore` exists it is the only
     *   default ignore.
     *
     * ```
     * # .usmignore
     * builddir/
     * *.sqlite
     * docs/generated/
     * ```
     */
    public class UsmIgnore : Object {

        private Vector<UsmIgnore.Pattern> patterns = new Vector<UsmIgnore.Pattern>();

        /**
         * Creates the default matcher: only the `.git` directory (and
         * everything beneath it) is ignored.
         */
        public UsmIgnore() {
            patterns.add(new UsmIgnore.Pattern(".git", true));
        }

        /**
         * Loads the `.usmignore` file at the root of {@link root_path},
         * falling back to the {@link UsmIgnore} defaults when the file is
         * absent.
         *
         * Throws when the file exists but cannot be read.
         */
        public UsmIgnore.from_root(string root_path) throws Error {
            this();

            var path = Path.build_filename(root_path, ".usmignore");
            if(!FileUtils.test(path, FileTest.EXISTS)) {
                return;
            }

            string contents;
            FileUtils.get_contents(path, out contents);
            foreach(var line in contents.split("\n")) {
                add_pattern(line);
            }
        }

        /**
         * Adds one pattern line; blank lines and `#` comments are skipped.
         */
        public void add_pattern(string line) {
            var trimmed = line.strip();
            if(trimmed.length == 0 || trimmed.has_prefix("#")) {
                return;
            }

            var directory_only = trimmed.has_suffix("/");
            var source = directory_only ? trimmed.substring(0, trimmed.length - 1) : trimmed;
            if(source.length == 0) {
                return;
            }
            patterns.add(new UsmIgnore.Pattern(source, directory_only));
        }

        /**
         * Whether the file or directory at {@link relative_path} (relative
         * to the package root, without a leading `./` or `/`) is ignored.
         */
        public bool matches(string relative_path, bool is_directory) {
            var path = relative_path.has_prefix("./") ? relative_path.substring(2) : relative_path;
            if(path == ".usmignore" || path == "MANIFEST.usm") {
                return false;
            }
            foreach(var pattern in patterns) {
                if(pattern.matches(path, is_directory)) {
                    return true;
                }
            }
            return false;
        }

        /**
         * Classic two-pointer wildcard match where `*` and `?` never match
         * the path separator.
         */
        private static bool glob_matches(string pattern, string text) {
            int pattern_index = 0;
            int text_index = 0;
            int star_index = -1;
            int backtrack_index = 0;

            while(text_index < text.length) {
                char pattern_char = pattern_index < pattern.length ? pattern[pattern_index] : '\0';
                char text_char = text[text_index];
                if((pattern_char == '?' && text_char != '/') || pattern_char == text_char) {
                    pattern_index++;
                    text_index++;
                }
                else if(pattern_char == '*') {
                    star_index = pattern_index++;
                    backtrack_index = text_index;
                }
                else if(star_index >= 0) {
                    pattern_index = star_index + 1;
                    text_index = ++backtrack_index;
                }
                else {
                    return false;
                }
            }

            while(pattern_index < pattern.length && pattern[pattern_index] == '*') {
                pattern_index++;
            }
            return pattern_index == pattern.length;
        }

        /**
         * One compiled `.usmignore` line. A pattern whose source contains
         * `/` is anchored to the full relative path, otherwise it matches
         * any single path segment suffix.
         */
        private class Pattern : Object {

            public string source;
            public bool directory_only;
            public bool anchored {
                get {
                    return source.contains("/");
                }
            }

            public Pattern(string source, bool directory_only) {
                this.source = source;
                this.directory_only = directory_only;
            }

            public bool matches(string path, bool is_directory) {
                if(directory_only) {
                    if(is_directory && segment_matches(path)) {
                        return true;
                    }
                    return matches_beneath(path);
                }
                if(anchored) {
                    return UsmIgnore.glob_matches(source, path);
                }
                return UsmIgnore.glob_matches(source, Path.get_basename(path));
            }

            private bool segment_matches(string path) {
                return anchored ? UsmIgnore.glob_matches(source, path)
                    : UsmIgnore.glob_matches(source, Path.get_basename(path));
            }

            /**
             * Everything beneath a directory matching a directory-only
             * pattern is ignored, whether or not the walker pruned the
             * directory itself.
             */
            private bool matches_beneath(string path) {
                var separator = path.index_of("/");
                while(separator != -1) {
                    if(segment_matches(path.substring(0, separator))) {
                        return true;
                    }
                    separator = path.index_of("/", separator + 1);
                }
                return false;
            }
        }
    }
}