Adds UOF formats support

mmalecot · Jan 28, 2024 · 0991fae · 0991fae
1 parent bc68481
commit 0991fae
Show file tree

Hide file tree

Showing 10 changed files with 136 additions and 70 deletions.
diff --git a/README.md b/README.md
@@ -221,6 +221,7 @@ identification.
 - Sun XML Writer (SXW)
 - Sun XML Writer Global (SGW)
 - Sun XML Writer Template (STW)
+- Uniform Office Format Text (UOT)
 - WordPerfect Document (WPD)
 
 ### Executable
@@ -508,6 +509,7 @@ identification.
 - StarImpress (SDD)
 - Sun XML Impress (SXI)
 - Sun XML Impress Template (STI)
+- Uniform Office Format Presentation (UOP)
 - WordPerfect Presentations (SHW)
 
 ### ROM
@@ -538,6 +540,7 @@ identification.
 - StarCalc (SDC)
 - Sun XML Calc (SXC)
 - Sun XML Calc Template (STC)
+- Uniform Office Format Spreadsheet (UOS)
 
 ### Subtitle
 

diff --git a/fixtures/document/sample.uot b/fixtures/document/sample.uot
@@ -0,0 +1,2 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<uof:UOF xmlns:uof="http://schemas.uof.org/cn/2003/uof" xmlns:表="http://schemas.uof.org/cn/2003/uof-spreadsheet" xmlns:演="http://schemas.uof.org/cn/2003/uof-slideshow" xmlns:字="http://schemas.uof.org/cn/2003/uof-wordproc" xmlns:图="http://schemas.uof.org/cn/2003/graph" xmlns:dt="uuid:C2F41010-65B3-11d1-A29F-00AA00C14882" xmlns:html="http://www.w3.org/TR/REC-html40" xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" uof:language="cn" uof:locID="u0000" uof:version="1.0" uof:mimetype="vnd.uof.text">
diff --git a/fixtures/presentation/sample.uop b/fixtures/presentation/sample.uop
@@ -0,0 +1,2 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<uof:UOF xmlns:uof="http://schemas.uof.org/cn/2003/uof" xmlns:图="http://schemas.uof.org/cn/2003/graph" xmlns:字="http://schemas.uof.org/cn/2003/uof-wordproc" xmlns:演="http://schemas.uof.org/cn/2003/uof-slideshow" xmlns:表="http://schemas.uof.org/cn/2003/uof-spreadsheet" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" uof:language="cn" uof:version="1.0" uof:locID="u0000" uof:mimetype="vnd.uof.presentation">
diff --git a/fixtures/spreadsheet/sample.uos b/fixtures/spreadsheet/sample.uos
@@ -0,0 +1,2 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<uof:UOF xmlns:uof="http://schemas.uof.org/cn/2003/uof" xmlns:表="http://schemas.uof.org/cn/2003/uof-spreadsheet" xmlns:演="http://schemas.uof.org/cn/2003/uof-slideshow" xmlns:字="http://schemas.uof.org/cn/2003/uof-wordproc" xmlns:图="http://schemas.uof.org/cn/2003/graph" xmlns:dt="uuid:C2F41010-65B3-11d1-A29F-00AA00C14882" xmlns:html="http://www.w3.org/TR/REC-html40" xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" uof:language="cn" uof:locID="u0000" uof:version="1.0" uof:mimetype="vnd.uof.spreadsheet">
diff --git a/src/formats.rs b/src/formats.rs
@@ -2656,6 +2656,27 @@ formats! {
     extension = "mod"
     kind = Audio
 
+    format = UniformOfficeFormatPresentation
+    name = "Uniform Office Format Presentation"
+    short_name = "UOP"
+    media_type = "application/vnd.uof.presentation"
+    extension = "uop"
+    kind = Presentation
+
+    format = UniformOfficeFormatSpreadsheet
+    name = "Uniform Office Format Spreadsheet"
+    short_name = "UOS"
+    media_type = "application/vnd.uof.spreadsheet"
+    extension = "uos"
+    kind = Spreadsheet
+
+    format = UniformOfficeFormatText
+    name = "Uniform Office Format Text"
+    short_name = "UOT"
+    media_type = "application/vnd.uof.text"
+    extension = "uot"
+    kind = Document
+
     format = Universal3d
     name = "Universal 3D"
     short_name = "U3D"

diff --git a/src/lib.rs b/src/lib.rs
@@ -138,6 +138,9 @@ identification.
   * [Tiled Tileset XML (TSX)](`FileFormat::TiledTilesetXml`)
   * [Timed Text Markup Language (TTML)](`FileFormat::TimedTextMarkupLanguage`)
   * [Training Center XML (TCX)](`FileFormat::TrainingCenterXml`)
+  * [Uniform Office Format Presentation (UOP)](`FileFormat::UniformOfficeFormatPresentation`)
+  * [Uniform Office Format Spreadsheet (UOS)](`FileFormat::UniformOfficeFormatSpreadsheet`)
+  * [Uniform Office Format Text (UOT)](`FileFormat::UniformOfficeFormatText`)
   * [Universal Subtitle Format (USF)](`FileFormat::UniversalSubtitleFormat`)
   * [XML Localization Interchange File Format (XLIFF)](`FileFormat::XmlLocalizationInterchangeFileFormat`)
   * [XML Shareable Playlist Format (XSPF)](`FileFormat::XmlShareablePlaylistFormat`)

diff --git a/src/readers.rs b/src/readers.rs
@@ -568,8 +568,8 @@ impl crate::FileFormat {
             // Determines the start index for searching the buffer.
             let start = if total_nread == 0 { OVERLAP_SIZE } else { 0 };
 
-            // Checks if the buffer contains the AI file format marker.
-            if buf[start..OVERLAP_SIZE + nread].find(AI_MARKER).is_some() {
+            // Checks if the buffer holds the AI file format marker.
+            if buf[start..OVERLAP_SIZE + nread].holds(AI_MARKER) {
                 return Ok(Self::AdobeIllustratorArtwork);
             }
 
@@ -685,75 +685,81 @@ impl crate::FileFormat {
 
     /// Determines file format from a XML reader.
     #[cfg(feature = "reader-xml")]
-    pub(crate) fn from_xml_reader<R: Read + Seek>(reader: R) -> Result<Self> {
-        // Maximum number of lines that can be processed by the reader.
-        const LINE_LIMIT: usize = 8;
-
-        // Maximum number of bytes that can be processed by the reader (32 KB).
-        const READ_LIMIT: u64 = 32_768;
-
-        // Creates a buffered reader.
-        let mut reader = BufReader::new(reader);
-
-        // Rewinds to the beginning of the stream.
-        reader.rewind()?;
+    pub(crate) fn from_xml_reader<R: Read + Seek>(mut reader: R) -> Result<Self> {
+        // Rewinds to the beginning of the stream plus the size of the XML file format signature.
+        reader.seek(SeekFrom::Start(5))?;
 
-        // Searches the reader for lines indicating the presence of various file formats.
-        for result in reader.take(READ_LIMIT).lines().take(LINE_LIMIT) {
-            let line = result?;
-            if line.contains("<abiword template=\"false\"") {
-                return Ok(Self::Abiword);
-            } else if line.contains("<abiword template=\"true\"") {
-                return Ok(Self::AbiwordTemplate);
-            } else if line.contains("<amf") {
-                return Ok(Self::AdditiveManufacturingFormat);
-            } else if line.contains("<ASX") || line.contains("<asx") {
-                return Ok(Self::AdvancedStreamRedirector);
-            } else if line.contains("<feed") {
-                return Ok(Self::Atom);
-            } else if line.contains("<COLLADA") || line.contains("<collada") {
-                return Ok(Self::DigitalAssetExchange);
-            } else if line.contains("<mxfile") {
-                return Ok(Self::Drawio);
-            } else if line.contains("<X3D") || line.contains("<x3d") {
-                return Ok(Self::Extensible3d);
-            } else if line.contains("<xsl") {
-                return Ok(Self::ExtensibleStylesheetLanguageTransformations);
-            } else if line.contains("<FictionBook") {
-                return Ok(Self::Fictionbook);
-            } else if line.contains("<gml") {
-                return Ok(Self::GeographyMarkupLanguage);
-            } else if line.contains("<gpx") {
-                return Ok(Self::GpsExchangeFormat);
-            } else if line.contains("<kml") {
-                return Ok(Self::KeyholeMarkupLanguage);
-            } else if line.contains("<math") {
-                return Ok(Self::MathematicalMarkupLanguage);
-            } else if line.contains("<MPD") {
-                return Ok(Self::MpegDashManifest);
-            } else if line.contains("<score-partwise") {
-                return Ok(Self::Musicxml);
-            } else if line.contains("<rss") {
-                return Ok(Self::ReallySimpleSyndication);
-            } else if line.contains("<SVG") || line.contains("<svg") {
-                return Ok(Self::ScalableVectorGraphics);
-            } else if line.contains("<soap") {
-                return Ok(Self::SimpleObjectAccessProtocol);
-            } else if line.contains("<map") {
-                return Ok(Self::TiledMapXml);
-            } else if line.contains("<tileset") {
-                return Ok(Self::TiledTilesetXml);
-            } else if line.contains("<tt") && line.contains("xmlns=\"http://www.w3.org/ns/ttml\"") {
-                return Ok(Self::TimedTextMarkupLanguage);
-            } else if line.contains("<TrainingCenterDatabase") {
-                return Ok(Self::TrainingCenterXml);
-            } else if line.contains("<USFSubtitles") {
-                return Ok(Self::UniversalSubtitleFormat);
-            } else if line.contains("<xliff") {
-                return Ok(Self::XmlLocalizationInterchangeFileFormat);
-            } else if line.contains("<playlist") {
-                return Ok(Self::XmlShareablePlaylistFormat);
-            }
+        // Creates and fills a buffer.
+        let mut buf = [0; 8192];
+        let nread = reader.read(&mut buf)?;
+
+        // Checks if the buffer holds markers indicating the presence of various file formats.
+        if buf[..nread].holds("<abiword template=\"false\"") {
+            return Ok(Self::Abiword);
+        } else if buf[..nread].holds("<abiword template=\"true\"") {
+            return Ok(Self::AbiwordTemplate);
+        } else if buf[..nread].holds("<amf") {
+            return Ok(Self::AdditiveManufacturingFormat);
+        } else if buf[..nread].holds("<ASX") || buf[..nread].holds("<asx") {
+            return Ok(Self::AdvancedStreamRedirector);
+        } else if buf[..nread].holds("<feed") {
+            return Ok(Self::Atom);
+        } else if buf[..nread].holds("<COLLADA") || buf[..nread].holds("<collada") {
+            return Ok(Self::DigitalAssetExchange);
+        } else if buf[..nread].holds("<mxfile") {
+            return Ok(Self::Drawio);
+        } else if buf[..nread].holds("<X3D") || buf[..nread].holds("<x3d") {
+            return Ok(Self::Extensible3d);
+        } else if buf[..nread].holds("<xsl") {
+            return Ok(Self::ExtensibleStylesheetLanguageTransformations);
+        } else if buf[..nread].holds("<FictionBook") {
+            return Ok(Self::Fictionbook);
+        } else if buf[..nread].holds("<gml") {
+            return Ok(Self::GeographyMarkupLanguage);
+        } else if buf[..nread].holds("<gpx") {
+            return Ok(Self::GpsExchangeFormat);
+        } else if buf[..nread].holds("<kml") {
+            return Ok(Self::KeyholeMarkupLanguage);
+        } else if buf[..nread].holds("<math") {
+            return Ok(Self::MathematicalMarkupLanguage);
+        } else if buf[..nread].holds("<MPD") {
+            return Ok(Self::MpegDashManifest);
+        } else if buf[..nread].holds("<score-partwise") {
+            return Ok(Self::Musicxml);
+        } else if buf[..nread].holds("<rss") {
+            return Ok(Self::ReallySimpleSyndication);
+        } else if buf[..nread].holds("<SVG") || buf[..nread].holds("<svg") {
+            return Ok(Self::ScalableVectorGraphics);
+        } else if buf[..nread].holds("<soap") {
+            return Ok(Self::SimpleObjectAccessProtocol);
+        } else if buf[..nread].holds("<map") {
+            return Ok(Self::TiledMapXml);
+        } else if buf[..nread].holds("<tileset") {
+            return Ok(Self::TiledTilesetXml);
+        } else if buf[..nread].holds("<tt")
+            && buf[..nread].holds("xmlns=\"http://www.w3.org/ns/ttml\"")
+        {
+            return Ok(Self::TimedTextMarkupLanguage);
+        } else if buf[..nread].holds("<TrainingCenterDatabase") {
+            return Ok(Self::TrainingCenterXml);
+        } else if buf[..nread].holds("<uof:UOF")
+            & buf[..nread].holds("uof:mimetype=\"vnd.uof.presentation\"")
+        {
+            return Ok(Self::UniformOfficeFormatPresentation);
+        } else if buf[..nread].holds("<uof:UOF")
+            & buf[..nread].holds("uof:mimetype=\"vnd.uof.spreadsheet\"")
+        {
+            return Ok(Self::UniformOfficeFormatSpreadsheet);
+        } else if buf[..nread].holds("<uof:UOF")
+            & buf[..nread].holds("uof:mimetype=\"vnd.uof.text\"")
+        {
+            return Ok(Self::UniformOfficeFormatText);
+        } else if buf[..nread].holds("<USFSubtitles") {
+            return Ok(Self::UniversalSubtitleFormat);
+        } else if buf[..nread].holds("<xliff") {
+            return Ok(Self::XmlLocalizationInterchangeFileFormat);
+        } else if buf[..nread].holds("<playlist") {
+            return Ok(Self::XmlShareablePlaylistFormat);
         }
 
         // Returns the default value.
@@ -1126,6 +1132,12 @@ trait FindBytes: AsRef<[u8]> {
         None
     }
 
+    /// Returns `true` if the data holds the specified byte pattern.
+    #[inline]
+    fn holds<P: AsRef<[u8]>>(&self, pat: P) -> bool {
+        self.find(pat).is_some()
+    }
+
     /// Searches for the specified byte pattern and returns the index of the last occurrence.
     fn rfind<P: AsRef<[u8]>>(&self, pat: P) -> Option<usize> {
         // Retrieves references to data and pattern.

diff --git a/tests/document.rs b/tests/document.rs
@@ -165,6 +165,13 @@ fn test_sun_xml_writer_template() {
     assert_eq!(fmt, FileFormat::SunXmlWriterTemplate);
 }
 
+#[cfg(feature = "reader-xml")]
+#[test]
+fn test_uniform_office_format_text() {
+    let fmt = FileFormat::from_file("fixtures/document/sample.uot").unwrap();
+    assert_eq!(fmt, FileFormat::UniformOfficeFormatText);
+}
+
 #[test]
 fn test_wordperfect_document_1() {
     let fmt = FileFormat::from_file("fixtures/document/sample1.wpd").unwrap();

diff --git a/tests/presentation.rs b/tests/presentation.rs
@@ -62,6 +62,13 @@ fn test_sun_xml_impress_template() {
     assert_eq!(fmt, FileFormat::SunXmlImpressTemplate);
 }
 
+#[cfg(feature = "reader-xml")]
+#[test]
+fn test_uniform_office_format_presentation() {
+    let fmt = FileFormat::from_file("fixtures/presentation/sample.uop").unwrap();
+    assert_eq!(fmt, FileFormat::UniformOfficeFormatPresentation);
+}
+
 #[test]
 fn test_wordperfect_presentations() {
     let fmt = FileFormat::from_file("fixtures/presentation/sample3.shw").unwrap();

diff --git a/tests/spreadsheet.rs b/tests/spreadsheet.rs
@@ -61,3 +61,10 @@ fn test_sun_xml_calc_template() {
     let fmt = FileFormat::from_file("fixtures/spreadsheet/sample.stc").unwrap();
     assert_eq!(fmt, FileFormat::SunXmlCalcTemplate);
 }
+
+#[cfg(feature = "reader-xml")]
+#[test]
+fn test_uniform_office_format_spreadsheet() {
+    let fmt = FileFormat::from_file("fixtures/spreadsheet/sample.uos").unwrap();
+    assert_eq!(fmt, FileFormat::UniformOfficeFormatSpreadsheet);
+}