fix: bom file

bokuweb · Feb 13, 2020 · f405404 · f405404
1 parent a49a343
commit f405404
Show file tree

Hide file tree

Showing 21 changed files with 141 additions and 18 deletions.
diff --git a/docx-core/examples/reader.rs b/docx-core/examples/reader.rs
@@ -3,7 +3,7 @@ use std::fs::*;
 use std::io::Read;
 
 pub fn main() {
-    let mut file = File::open("./run.docx").unwrap();
+    let mut file = File::open("./10.docx").unwrap();
     let mut buf = vec![];
     file.read_to_end(&mut buf).unwrap();
     dbg!(read_docx(&buf).unwrap().json());

diff --git a/docx-core/src/documents/content_types.rs b/docx-core/src/documents/content_types.rs
@@ -111,7 +111,7 @@ impl FromXML for ContentTypes {
                 Ok(XmlEvent::EndElement { .. }) => {
                     depth -= 1;
                 }
-                Err(_) => return Err(ReaderError::XMLReadError),
+                Err(_) => {}
                 _ => {}
             }
         }

diff --git a/docx-core/src/reader/document_rels.rs b/docx-core/src/reader/document_rels.rs
@@ -29,8 +29,8 @@ pub fn read_document_rels(
         .ok_or(ReaderError::DocumentRelsNotFoundError)?;
     let p = find_rels_filename(&main_path)?;
     let p = p.to_str().ok_or(ReaderError::DocumentRelsNotFoundError)?;
-    let rels_xml = archive.by_name(&p)?;
-    let rels = read_rels_xml(rels_xml, dir)?;
+    let data = read_zip(archive, &p)?;
+    let rels = read_rels_xml(&data[..], dir)?;
     Ok(rels)
 }
 

diff --git a/docx-core/src/reader/mod.rs b/docx-core/src/reader/mod.rs
@@ -9,6 +9,7 @@ mod level;
 mod numbering_property;
 mod numberings;
 mod paragraph;
+mod read_zip;
 mod rels;
 mod run;
 mod style;
@@ -27,6 +28,7 @@ pub use attributes::*;
 pub use document_rels::*;
 pub use errors::ReaderError;
 pub use from_xml::*;
+pub use read_zip::*;
 pub use xml_element::*;
 
 const DOC_RELATIONSHIP_TYPE: &str =
@@ -41,36 +43,49 @@ pub fn read_docx(buf: &[u8]) -> Result<Docx, ReaderError> {
     let mut archive = zip::ZipArchive::new(cur)?;
     // First, the content type for relationship parts and the Main Document part
     // (the only required part) must be defined (physically located at /[Content_Types].xml in the package)
-    let content_types_xml = archive.by_name("[Content_Types].xml")?;
-    let _content_types = ContentTypes::from_xml(content_types_xml)?;
+    let _content_types = {
+        let data = read_zip(&mut archive, "[Content_Types].xml")?;
+        ContentTypes::from_xml(&data[..])?
+    };
+
     // Next, the single required relationship (the package-level relationship to the Main Document part)
     //  must be defined (physically located at /_rels/.rels in the package)
-    let rels_xml = archive.by_name("_rels/.rels")?;
-    let rels = Rels::from_xml(rels_xml)?;
+    let rels = {
+        let data = read_zip(&mut archive, "_rels/.rels")?;
+        Rels::from_xml(&data[..])?
+    };
     // Finally, the minimum content for the Main Document part must be defined
     // (physically located at /document.xml in the package):
     let main_rel = rels
         .find_target(DOC_RELATIONSHIP_TYPE)
         .ok_or(ReaderError::DocumentNotFoundError)?;
-    let document_xml = archive.by_name(&main_rel.2)?;
-    let document = Document::from_xml(document_xml)?;
+    let document = {
+        let data = read_zip(&mut archive, &main_rel.2)?;
+        Document::from_xml(&data[..])?
+    };
     let mut docx = Docx::new().document(document);
     // Read document relationships
     let rels = read_document_rels(&mut archive, &main_rel.2)?;
 
     // Read styles
     let style_path = rels.find_target_path(STYLE_RELATIONSHIP_TYPE);
     if let Some(style_path) = style_path {
-        let styles_xml = archive.by_name(style_path.to_str().expect("should have styles"))?;
-        let styles = Styles::from_xml(styles_xml)?;
+        let data = read_zip(
+            &mut archive,
+            style_path.to_str().expect("should have styles"),
+        )?;
+        let styles = Styles::from_xml(&data[..])?;
         docx = docx.styles(styles);
     }
 
     // Read numberings
     let num_path = rels.find_target_path(NUMBERING_RELATIONSHIP_TYPE);
     if let Some(num_path) = num_path {
-        let num_xml = archive.by_name(num_path.to_str().expect("should have numberings"))?;
-        let nums = Numberings::from_xml(num_xml)?;
+        let data = read_zip(
+            &mut archive,
+            num_path.to_str().expect("should have numberings"),
+        )?;
+        let nums = Numberings::from_xml(&data[..])?;
         docx = docx.numberings(nums);
     }
 

diff --git a/docx-core/src/reader/read_zip.rs b/docx-core/src/reader/read_zip.rs
@@ -0,0 +1,24 @@
+use std::io::{Cursor, Read};
+use zip;
+
+use super::ReaderError;
+
+pub fn read_zip(
+    archive: &mut zip::read::ZipArchive<Cursor<&[u8]>>,
+    name: &str,
+) -> Result<Vec<u8>, ReaderError> {
+    let mut p = name.to_owned();
+    if p.starts_with("/") {
+        p.remove(0);
+    }
+    let mut xml = archive.by_name(&p)?;
+    let mut data = vec![];
+    xml.read_to_end(&mut data).unwrap();
+    // Remove BOM
+    if (data[0] == 0xef) && (data[1] == 0xbb) && (data[2] == 0xbf) {
+        data.remove(0);
+        data.remove(0);
+        data.remove(0);
+    }
+    Ok(data)
+}
diff --git a/docx-core/src/reader/styles.rs b/docx-core/src/reader/styles.rs
@@ -47,8 +47,7 @@ mod tests {
 
     #[test]
     fn test_from_xml() {
-        let xml =
-            r#"<w:styles xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let xml = r#"<w:styles xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
     <w:style w:type="character" w:styleId="FootnoteTextChar">
         <w:name w:val="Footnote Text Char"></w:name>
         <w:rPr>

diff --git a/docx-core/src/types/alignment_type.rs b/docx-core/src/types/alignment_type.rs
@@ -10,7 +10,9 @@ pub enum AlignmentType {
     Center,
     Left,
     Right,
+    Both,
     Justified,
+    Unsupported,
 }
 
 impl fmt::Display for AlignmentType {
@@ -19,7 +21,9 @@ impl fmt::Display for AlignmentType {
             AlignmentType::Center => write!(f, "center"),
             AlignmentType::Left => write!(f, "left"),
             AlignmentType::Right => write!(f, "right"),
+            AlignmentType::Both => write!(f, "both"),
             AlignmentType::Justified => write!(f, "justified"),
+            _ => write!(f, "unsupported"),
         }
     }
 }
@@ -31,8 +35,9 @@ impl FromStr for AlignmentType {
             "left" => Ok(AlignmentType::Left),
             "right" => Ok(AlignmentType::Right),
             "center" => Ok(AlignmentType::Center),
+            "both" => Ok(AlignmentType::Both),
             "justified" => Ok(AlignmentType::Justified),
-            _ => Err(errors::TypeError::FromStrError),
+            _ => Ok(AlignmentType::Unsupported),
         }
     }
 }
diff --git a/docx-core/src/types/style_type.rs b/docx-core/src/types/style_type.rs
@@ -12,6 +12,7 @@ pub enum StyleType {
     Paragraph,
     Character,
     Numbering,
+    Table,
     Unsupported,
 }
 
@@ -21,6 +22,7 @@ impl fmt::Display for StyleType {
             StyleType::Paragraph => write!(f, "paragraph"),
             StyleType::Character => write!(f, "character"),
             StyleType::Numbering => write!(f, "numbering"),
+            StyleType::Table => write!(f, "table"),
             StyleType::Unsupported => write!(f, "unsupported"),
         }
     }
@@ -33,6 +35,7 @@ impl FromStr for StyleType {
             "paragraph" => Ok(StyleType::Paragraph),
             "character" => Ok(StyleType::Character),
             "numbering" => Ok(StyleType::Numbering),
+            "table" => Ok(StyleType::Table),
             _ => Ok(StyleType::Unsupported),
         }
     }

diff --git a/docx-core/tests/reader.rs b/docx-core/tests/reader.rs
@@ -139,3 +139,18 @@ pub fn read_table_merged_libre_office() {
     file.write_all(json.as_bytes()).unwrap();
     file.flush().unwrap();
 }
+
+#[test]
+pub fn read_bom() {
+    let mut file = File::open("../fixtures/bom/bom.docx").unwrap();
+    let mut buf = vec![];
+    file.read_to_end(&mut buf).unwrap();
+    let json = read_docx(&buf).unwrap().json();
+
+    assert_debug_snapshot!(&json);
+
+    let path = std::path::Path::new("./tests/output/bom.json");
+    let mut file = std::fs::File::create(&path).unwrap();
+    file.write_all(json.as_bytes()).unwrap();
+    file.flush().unwrap();
+}
diff --git a/docx-core/tests/snapshots/lib__reader__read_bom.snap b/docx-core/tests/snapshots/lib__reader__read_bom.snap
@@ -0,0 +1,5 @@
+---
+source: docx-core/tests/reader.rs
+expression: "&json"
+---
+"{\n  \"contentType\": {\n    \"types\": {\n      \"/_rels/.rels\": \"application/vnd.openxmlformats-package.relationships+xml\",\n      \"/docProps/app.xml\": \"application/vnd.openxmlformats-officedocument.extended-properties+xml\",\n      \"/docProps/core.xml\": \"application/vnd.openxmlformats-package.core-properties+xml\",\n      \"/word/_rels/document.xml.rels\": \"application/vnd.openxmlformats-package.relationships+xml\",\n      \"/word/comments.xml\": \"application/vnd.openxmlformats-officedocument.wordprocessingml.comments+xml\",\n      \"/word/document.xml\": \"application/vnd.openxmlformats-officedocument.wordprocessingml.document.main+xml\",\n      \"/word/fontTable.xml\": \"application/vnd.openxmlformats-officedocument.wordprocessingml.fontTable+xml\",\n      \"/word/numbering.xml\": \"application/vnd.openxmlformats-officedocument.wordprocessingml.numbering+xml\",\n      \"/word/settings.xml\": \"application/vnd.openxmlformats-officedocument.wordprocessingml.settings+xml\",\n      \"/word/styles.xml\": \"application/vnd.openxmlformats-officedocument.wordprocessingml.styles+xml\"\n    }\n  },\n  \"rels\": {\n    \"rels\": [\n      [\n        \"http://schemas.openxmlformats.org/package/2006/relationships/metadata/core-properties\",\n        \"rId1\",\n        \"docProps/core.xml\"\n      ],\n      [\n        \"http://schemas.openxmlformats.org/officeDocument/2006/relationships/extended-properties\",\n        \"rId2\",\n        \"docProps/app.xml\"\n      ],\n      [\n        \"http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument\",\n        \"rId3\",\n        \"word/document.xml\"\n      ]\n    ]\n  },\n  \"documentRels\": {\n    \"hasComments\": false,\n    \"hasNumberings\": false\n  },\n  \"docProps\": {\n    \"app\": {},\n    \"core\": {\n      \"config\": {\n        \"created\": null,\n        \"creator\": null,\n        \"description\": null,\n        \"language\": null,\n        \"lastModifiedBy\": null,\n        \"modified\": null,\n        \"revision\": null,\n        \"subject\": null,\n        \"title\": null\n      }\n    }\n  },\n  \"styles\": {\n    \"docDefaults\": {\n      \"runPropertyDefault\": {\n        \"runProperty\": {\n          \"sz\": null,\n          \"szCs\": null,\n          \"color\": null,\n          \"highlight\": null,\n          \"underline\": null,\n          \"bold\": null,\n          \"boldCs\": null,\n          \"italic\": null,\n          \"italicCs\": null,\n          \"vanish\": null\n        }\n      }\n    },\n    \"styles\": [\n      {\n        \"styleId\": \"Normal\",\n        \"name\": \"Normal\",\n        \"styleType\": \"paragraph\",\n        \"runProperty\": {\n          \"sz\": null,\n          \"szCs\": null,\n          \"color\": null,\n          \"highlight\": null,\n          \"underline\": null,\n          \"bold\": null,\n          \"boldCs\": null,\n          \"italic\": null,\n          \"italicCs\": null,\n          \"vanish\": null\n        },\n        \"paragraphProperty\": {\n          \"runProperty\": {\n            \"sz\": null,\n            \"szCs\": null,\n            \"color\": null,\n            \"highlight\": null,\n            \"underline\": null,\n            \"bold\": null,\n            \"boldCs\": null,\n            \"italic\": null,\n            \"italicCs\": null,\n            \"vanish\": null\n          },\n          \"style\": \"Normal\",\n          \"numberingProperty\": null,\n          \"alignment\": \"both\",\n          \"indent\": null\n        }\n      },\n      {\n        \"styleId\": \"DefaultParagraphFont\",\n        \"name\": \"Default Paragraph Font\",\n        \"styleType\": \"character\",\n        \"runProperty\": {\n          \"sz\": null,\n          \"szCs\": null,\n          \"color\": null,\n          \"highlight\": null,\n          \"underline\": null,\n          \"bold\": null,\n          \"boldCs\": null,\n          \"italic\": null,\n          \"italicCs\": null,\n          \"vanish\": null\n        },\n        \"paragraphProperty\": {\n          \"runProperty\": {\n            \"sz\": null,\n            \"szCs\": null,\n            \"color\": null,\n            \"highlight\": null,\n            \"underline\": null,\n            \"bold\": null,\n            \"boldCs\": null,\n            \"italic\": null,\n            \"italicCs\": null,\n            \"vanish\": null\n          },\n          \"style\": \"Normal\",\n          \"numberingProperty\": null,\n          \"alignment\": null,\n          \"indent\": null\n        }\n      },\n      {\n        \"styleId\": \"TableNormal\",\n        \"name\": \"Normal Table\",\n        \"styleType\": \"table\",\n        \"runProperty\": {\n          \"sz\": null,\n          \"szCs\": null,\n          \"color\": null,\n          \"highlight\": null,\n          \"underline\": null,\n          \"bold\": null,\n          \"boldCs\": null,\n          \"italic\": null,\n          \"italicCs\": null,\n          \"vanish\": null\n        },\n        \"paragraphProperty\": {\n          \"runProperty\": {\n            \"sz\": null,\n            \"szCs\": null,\n            \"color\": null,\n            \"highlight\": null,\n            \"underline\": null,\n            \"bold\": null,\n            \"boldCs\": null,\n            \"italic\": null,\n            \"italicCs\": null,\n            \"vanish\": null\n          },\n          \"style\": \"Normal\",\n          \"numberingProperty\": null,\n          \"alignment\": null,\n          \"indent\": null\n        }\n      },\n      {\n        \"styleId\": \"NoList\",\n        \"name\": \"No List\",\n        \"styleType\": \"numbering\",\n        \"runProperty\": {\n          \"sz\": null,\n          \"szCs\": null,\n          \"color\": null,\n          \"highlight\": null,\n          \"underline\": null,\n          \"bold\": null,\n          \"boldCs\": null,\n          \"italic\": null,\n          \"italicCs\": null,\n          \"vanish\": null\n        },\n        \"paragraphProperty\": {\n          \"runProperty\": {\n            \"sz\": null,\n            \"szCs\": null,\n            \"color\": null,\n            \"highlight\": null,\n            \"underline\": null,\n            \"bold\": null,\n            \"boldCs\": null,\n            \"italic\": null,\n            \"italicCs\": null,\n            \"vanish\": null\n          },\n          \"style\": \"Normal\",\n          \"numberingProperty\": null,\n          \"alignment\": null,\n          \"indent\": null\n        }\n      }\n    ]\n  },\n  \"document\": {\n    \"children\": [\n      {\n        \"type\": \"paragraph\",\n        \"data\": {\n          \"children\": [\n            {\n              \"type\": \"unsupported\"\n            },\n            {\n              \"type\": \"unsupported\"\n            },\n            {\n              \"type\": \"run\",\n              \"data\": {\n                \"runProperty\": {\n                  \"sz\": null,\n                  \"szCs\": null,\n                  \"color\": null,\n                  \"highlight\": null,\n                  \"underline\": null,\n                  \"bold\": null,\n                  \"boldCs\": null,\n                  \"italic\": null,\n                  \"italicCs\": null,\n                  \"vanish\": null\n                },\n                \"children\": [\n                  {\n                    \"type\": \"text\",\n                    \"data\": {\n                      \"preserveSpace\": true,\n                      \"text\": \"Hello\"\n                    }\n                  }\n                ]\n              }\n            }\n          ],\n          \"property\": {\n            \"runProperty\": {\n              \"sz\": null,\n              \"szCs\": null,\n              \"color\": null,\n              \"highlight\": null,\n              \"underline\": null,\n              \"bold\": null,\n              \"boldCs\": null,\n              \"italic\": null,\n              \"italicCs\": null,\n              \"vanish\": null\n            },\n            \"style\": \"Normal\",\n            \"numberingProperty\": null,\n            \"alignment\": null,\n            \"indent\": null\n          },\n          \"hasNumbering\": false,\n          \"attrs\": []\n        }\n      }\n    ],\n    \"sectionProperty\": {\n      \"pageSize\": {\n        \"w\": 11906,\n        \"h\": 16838\n      },\n      \"pageMargin\": {\n        \"top\": 1985,\n        \"left\": 1701,\n        \"bottom\": 1701,\n        \"right\": 1701,\n        \"header\": 851,\n        \"footer\": 992,\n        \"gutter\": 0\n      },\n      \"columns\": 425,\n      \"documentGrid\": 360\n    },\n    \"hasNumbering\": false\n  },\n  \"comments\": {\n    \"comments\": []\n  },\n  \"numberings\": {\n    \"abstractNums\": [],\n    \"numberings\": []\n  },\n  \"settings\": {\n    \"defaultTabStop\": 709,\n    \"zoom\": 100\n  },\n  \"fontTable\": {}\n}"
diff --git a/docx-core/tests/snapshots/lib__reader__read_indent_word_online.snap b/docx-core/tests/snapshots/lib__reader__read_indent_word_online.snap