feat(rg): add max-filesize filtering

crates/bashkit/src/builtins/rg/mod.rs

@@ -56,6 +56,7 @@ struct RgOptions {
     max_count: Option<usize>,
     max_columns: Option<usize>,
     max_columns_preview: bool,
+    max_filesize: Option<u64>,
     max_depth: Option<usize>,
     before_context: usize,
     after_context: usize,
@@ -179,6 +180,7 @@ impl RgOptions {
             max_count: None,
             max_columns: None,
             max_columns_preview: false,
+            max_filesize: None,
             max_depth: None,
             before_context: 0,
             after_context: 0,
@@ -254,6 +256,8 @@ impl RgOptions {
                 opts.max_columns = Some(val.parse().map_err(|_| {
                     Error::Execution(format!("rg: invalid --max-columns value: {val}"))
                 })?);
+            } else if let Some(val) = long_value(&mut p, "--max-filesize")? {
+                opts.max_filesize = Some(parse_max_filesize(&val)?);
             } else if let Some(val) = long_value(&mut p, "--max-depth")? {
                 opts.max_depth = Some(val.parse().map_err(|_| {
                     Error::Execution(format!("rg: invalid --max-depth value: {val}"))
@@ -815,6 +819,10 @@ impl RgOptions {
             .any(|file_type| file_type.matches(path))
     }
 
+    fn matches_max_filesize(&self, size: u64) -> bool {
+        self.max_filesize.is_none_or(|max| size <= max)
+    }
+
     fn first_positive_glob(&self) -> Option<String> {
         self.glob_rules
             .iter()
@@ -1211,6 +1219,31 @@ fn parse_context_value(value: &str, flag: &str) -> Result<usize> {
         .map_err(|_| Error::Execution(format!("rg: invalid {} value: {}", flag, value)))
 }
 
+fn parse_max_filesize(value: &str) -> Result<u64> {
+    let (digits, multiplier) = match value.as_bytes().last().copied() {
+        Some(b'K') => (&value[..value.len() - 1], 1024_u64),
+        Some(b'M') => (&value[..value.len() - 1], 1024_u64 * 1024),
+        Some(b'G') => (&value[..value.len() - 1], 1024_u64 * 1024 * 1024),
+        _ => (value, 1),
+    };
+
+    if digits.is_empty() || !digits.bytes().all(|b| b.is_ascii_digit()) {
+        return Err(Error::Execution(format!(
+            "rg: error parsing flag --max-filesize: invalid size: invalid format for size '{value}', which should be a non-empty sequence of digits followed by an optional 'K', 'M' or 'G' suffix"
+        )));
+    }
+
+    digits
+        .parse::<u64>()
+        .ok()
+        .and_then(|size| size.checked_mul(multiplier))
+        .ok_or_else(|| {
+            Error::Execution(format!(
+                "rg: error parsing flag --max-filesize: invalid size: size '{value}' is too large"
+            ))
+        })
+}
+
 fn parse_path_separator(value: &str) -> Result<String> {
     if value.len() == 1 {
         Ok(value.to_string())
@@ -1558,6 +1591,7 @@ async fn collect_rg_files_recursive(
                     && opts
                         .max_depth
                         .is_none_or(|max_depth| entry_depth <= max_depth)
+                    && opts.matches_max_filesize(entry.metadata.size)
                     && !opts.is_ignored_by_rules(&path, false, &rules)
                     && opts.matches_globs(&path, cwd)
                     && opts.matches_type_filters(&path)
@@ -1567,7 +1601,9 @@ async fn collect_rg_files_recursive(
             }
         } else if let Ok(meta) = fs.stat(&current).await
             && meta.file_type.is_file()
+            && opts.matches_max_filesize(meta.size)
             && opts.matches_globs(&current, cwd)
+            && opts.matches_type_filters(&current)
         {
             result.push(current);
         }
@@ -1661,6 +1697,7 @@ async fn try_indexed_search(
         || opts.files_without_matches
         || opts.uses_ignore_files()
         || opts.patterns.len() != 1
+        || opts.max_filesize.is_some()
         || !opts.type_includes.is_empty()
         || !opts.type_excludes.is_empty()
     {
@@ -2171,6 +2208,10 @@ fn append_rg_stats(output: &mut String, stats: &RgSearchStats, bytes_printed: us
 impl Builtin for Rg {
     async fn execute(&self, ctx: Context<'_>) -> Result<ExecResult> {
         let help_text = "Usage: rg [OPTIONS] PATTERN [PATH...]\nRecursively search for a pattern.\n\n  -i, --ignore-case\tcase insensitive\n  -S, --smart-case\tcase insensitive if pattern is lowercase\n  -s, --case-sensitive\tcase sensitive\n  -n, --line-number\tshow line numbers\n  -N, --no-line-number\tsuppress line numbers\n  --column\tshow column numbers\n  -b, --byte-offset\tshow byte offsets\n  --vimgrep\tshow file:line:column:match lines\n  --json\tshow JSON Lines events\n  --stats\tshow search statistics\n  --null\tterminate path fields with NUL\n  -c, --count\tcount matching lines\n  --count-matches\tcount individual matches\n  --include-zero\tinclude zero counts\n  -l, --files-with-matches\tfiles with matches\n  --files-without-match\tfiles without matches\n  --files\tprint files that would be searched\n  -v, --invert-match\tinvert match\n  -w, --word-regexp\tword boundary\n  -x, --line-regexp\tmatch whole lines\n  -F, --fixed-strings\tfixed strings (literal)\n  -a, --text\tsearch binary files as text\n  --binary\tsearch binary files and print binary-match summaries\n  --crlf\ttreat CRLF as line terminators for $ anchors\n  --no-crlf\tdisable CRLF line terminator mode\n  -U, --multiline\tenable matching across line boundaries\n  --no-multiline\tdisable multiline matching\n  --multiline-dotall\tmake . match line terminators in multiline mode\n  --no-multiline-dotall\tdisable multiline dotall mode\n  -o, --only-matching\tshow only matching text\n  -q, --quiet\tsuppress output; exit status only\n  -e, --regexp PATTERN\tuse PATTERN for matching\n  -f, --file PATTERNFILE\tread patterns from file\n  -E, --encoding ENCODING\tdecode searched files using ENCODING\n  -r, --replace REPLACEMENT\treplace matches in output\n  --passthru\tprint matching and non-matching lines\n  --trim\ttrim whitespace from output lines\n  -m, --max-count NUM\tmax count per file\n  -M, --max-columns NUM\tomit lines longer than NUM columns\n  --max-columns-preview\tshow prefixes of long lines\n  --max-depth NUM\tlimit recursive directory depth\n  -A, --after-context NUM\tshow trailing context\n  -B, --before-context NUM\tshow leading context\n  -C, --context NUM\tshow leading and trailing context\n  --context-separator SEP\tset context group separator\n  --field-match-separator SEP\tset match field separator\n  --field-context-separator SEP\tset context field separator\n  --heading\tgroup matches by file\n  --no-heading\tdisable heading output\n  --sort SORTBY\tsort paths (path only)\n  --sortr SORTBY\tsort paths in reverse (path only)\n  --sort-files\tsort --files output\n  --path-separator SEP\tset displayed path separator\n  -g, --glob GLOB\tinclude/exclude paths by glob (!GLOB excludes)\n  -t, --type TYPE\tinclude files matching TYPE\n  -T, --type-not TYPE\texclude files matching TYPE\n  --type-add TYPE:GLOB\tadd a file type glob\n  --type-clear TYPE\tclear a file type definition\n  --type-list\tshow file type definitions\n  --ignore-file FILE\tuse additional ignore file\n  --no-ignore\tdo not use ignore files\n  --no-ignore-dot\tdo not use .ignore files\n  --no-ignore-vcs\tdo not use .gitignore files\n  --no-require-git\tuse .gitignore outside git repositories\n  --require-git\trequire a git repository for .gitignore files\n  -u, --unrestricted\treduce filtering (repeatable)\n  --messages\tshow file read diagnostics\n  --no-messages\tsuppress file read diagnostics\n  --hidden\tsearch hidden files and directories\n  --no-hidden\tdo not search hidden files and directories\n  -H, --with-filename\tshow filename\n  -I, --no-filename\tsuppress filename\n  --line-buffered\tforce line buffering (no-op)\n  --block-buffered\tforce block buffering (no-op)\n  --no-config\tdo not read config files (no-op)\n  --mmap\tsearch using memory maps when possible (no-op)\n  --no-mmap\tdisable memory maps (no-op)\n  -P, --pcre2\tuse PCRE2 regex engine for supported patterns (no-op)\n  --no-pcre2\tdisable PCRE2 regex engine (no-op)\n  --engine ENGINE\tselect regex engine: default, auto, pcre2 (no-op)\n  --auto-hybrid-regex\tuse PCRE2 when needed (no-op)\n  --no-auto-hybrid-regex\tdisable auto hybrid regex (no-op)\n  --color MODE\tcolor output (no-op)\n  -h, --help\tdisplay this help and exit\n  -V, --version\toutput version information and exit\n";
+        let help_text = help_text.replace(
+            "  --max-columns-preview\tshow prefixes of long lines\n",
+            "  --max-columns-preview\tshow prefixes of long lines\n  --max-filesize NUM\tignore files larger than NUM bytes, K, M, or G\n",
+        );
         let help_text = help_text.replace(
             "  -g, --glob GLOB\tinclude/exclude paths by glob (!GLOB excludes)\n",
             "  -g, --glob GLOB\tinclude/exclude paths by glob (!GLOB excludes)\n  --iglob GLOB\tcase-insensitive include/exclude path glob\n  --glob-case-insensitive\tmake -g/--glob rules case-insensitive\n  --no-glob-case-insensitive\tdisable case-insensitive -g/--glob rules\n",
@@ -3201,6 +3242,11 @@ mod tests {
         b"short needle\ncontext line is long\n0123456789 needle\nnomatch\n",
     )];
 
+    const DIFF_MAX_FILESIZE_FILES: &[(&str, &[u8])] = &[
+        ("/proj/small.txt", b"needle\n"),
+        ("/proj/big.txt", b"needle long\n"),
+    ];
+
     const RG_DIFF_CASES: &[RgDiffCase] = &[
         RgDiffCase {
             name: "stdin line numbers",
@@ -4270,6 +4316,52 @@ mod tests {
             cwd: "/",
             output: RgDiffOutput::Exact,
         },
+        RgDiffCase {
+            name: "max filesize filters recursive search",
+            args: &["--max-filesize", "10", "needle", "proj"],
+            stdin: None,
+            files: DIFF_MAX_FILESIZE_FILES,
+            cwd: "/",
+            output: RgDiffOutput::Exact,
+        },
+        RgDiffCase {
+            name: "max filesize suffix allows larger files",
+            args: &["--max-filesize", "1K", "needle", "proj"],
+            stdin: None,
+            files: DIFF_MAX_FILESIZE_FILES,
+            cwd: "/",
+            output: RgDiffOutput::UnorderedLines,
+        },
+        RgDiffCase {
+            name: "max filesize does not filter explicit file",
+            args: &[
+                "--max-filesize",
+                "10",
+                "needle",
+                "proj/small.txt",
+                "proj/big.txt",
+            ],
+            stdin: None,
+            files: DIFF_MAX_FILESIZE_FILES,
+            cwd: "/",
+            output: RgDiffOutput::Exact,
+        },
+        RgDiffCase {
+            name: "max filesize filters files listing",
+            args: &["--files", "--max-filesize", "10", "proj"],
+            stdin: None,
+            files: DIFF_MAX_FILESIZE_FILES,
+            cwd: "/",
+            output: RgDiffOutput::Exact,
+        },
+        RgDiffCase {
+            name: "max filesize empty files listing exits one",
+            args: &["--files", "--max-filesize", "1", "proj"],
+            stdin: None,
+            files: DIFF_MAX_FILESIZE_FILES,
+            cwd: "/",
+            output: RgDiffOutput::Exact,
+        },
         RgDiffCase {
             name: "missing file keeps stdout and exits 2",
             args: &["needle", "proj/a.txt", "proj/missing.txt"],