feat(html): extract HTML meta, Dublin Core and Office tags

src/core/file_format.rs

@@ -249,6 +249,9 @@ pub enum FileFormat {
     /// Plain text format (.txt)
     TXT,
 
+    /// HTML / XHTML document (.htm, .html, .xhtml)
+    HTML,
+
     /// Windows shortcut (.lnk)
     LNK,
 
@@ -378,6 +381,7 @@ impl FileFormat {
             FileFormat::ICS => "iCalendar",
             FileFormat::EML => "EML",
             FileFormat::TXT => "TXT",
+            FileFormat::HTML => "HTML",
             FileFormat::LNK => "Windows Shortcut",
             FileFormat::SQLite => "SQLite",
             FileFormat::Prefetch => "Windows Prefetch",
@@ -478,6 +482,7 @@ impl FileFormat {
             FileFormat::ICS => &["ics", "ical"],
             FileFormat::EML => &["eml", "email"],
             FileFormat::TXT => &["txt", "text"],
+            FileFormat::HTML => &["htm", "html", "xhtml"],
             FileFormat::LNK => &["lnk"],
             FileFormat::SQLite => &["db", "sqlite", "sqlite3"],
             FileFormat::Prefetch => &["pf"],

src/core/format_dispatch.rs

@@ -68,6 +68,7 @@ use crate::parsers::specialized::sqlite::parse_sqlite_metadata;
 use crate::parsers::specialized::stl::parse_stl_metadata;
 use crate::parsers::specialized::x509::parse_x509_metadata;
 use crate::parsers::text::eps::parse_eps_metadata;
+use crate::parsers::text::html::parse_html_metadata;
 use crate::parsers::text::txt::parse_txt_metadata;
 use crate::parsers::text::vcf::parse_vcf_metadata;
 use crate::parsers::video::asf::parse_asf_metadata;
@@ -170,6 +171,7 @@ pub fn dispatch_format_parser(reader: &dyn FileReader, format: FileFormat) -> Re
         FileFormat::HDF5 => convert_string_error(parse_hdf5_metadata(reader), "HDF5"),
         FileFormat::VCF => convert_string_error(parse_vcf_metadata(reader), "VCF"),
         FileFormat::TXT => convert_string_error(parse_txt_metadata(reader), "TXT"),
+        FileFormat::HTML => convert_string_error(parse_html_metadata(reader), "HTML"),
         FileFormat::LNK => convert_string_error(parse_lnk_metadata(reader), "LNK"),
         FileFormat::SQLite => convert_string_error(parse_sqlite_metadata(reader), "SQLite"),
         FileFormat::ICS => convert_string_error(parse_ics_metadata(reader), "ICS"),

src/parsers/detection/mod.rs

@@ -266,6 +266,18 @@ pub fn detect_format(reader: &dyn FileReader) -> io::Result<FileFormat> {
         return Ok(FileFormat::Plist);
     }
 
+    // HTML and XHTML, using ExifTool's own gate from `HTML.pm`'s ProcessHTML.
+    // It runs after the SVG and plist roots because those three share the
+    // `<?xml` opening: ExifTool requires an actual HTML element in the first
+    // 256 bytes before it will accept an XML declaration as HTML, which is
+    // what keeps SVG, XMP sidecars and XML plists out. Before this existed
+    // every .html file fell through to the plain-text fallback and was parsed
+    // as TXT, reporting five TEXT:* statistics ExifTool never reports and none
+    // of the 57 HTML/Dublin-Core/Office tags it does.
+    if crate::parsers::text::html::looks_like_html(magic_bytes) {
+        return Ok(FileFormat::HTML);
+    }
+
     // Text-based formats need a wider bounded probe for long ICS bodies and EML header blocks.
     let text_probe_len = reader
         .size()

src/parsers/font/ttf.rs

@@ -195,7 +195,7 @@ impl TTFParser {
     }
 
     /// Decodes a string using the Macintosh Roman encoding.
-    fn decode_mac_roman(data: &[u8]) -> String {
+    pub(crate) fn decode_mac_roman(data: &[u8]) -> String {
         const MAC_ROMAN_HIGH: [char; 128] = [
             'Ä', 'Å', 'Ç', 'É', 'Ñ', 'Ö', 'Ü', 'á', 'à', 'â', 'ä', 'ã', 'å', 'ç', 'é', 'è', 'ê',
             'ë', 'í', 'ì', 'î', 'ï', 'ñ', 'ó', 'ò', 'ô', 'ö', 'õ', 'ú', 'ù', 'û', 'ü', '†', '°',