Add json definitions

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());

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(_) => {}
                 _ => {}
             }
         }

docx-core/src/documents/elements/paragraph.rs

@@ -69,6 +69,25 @@ impl Serialize for ParagraphChild {
                 t.serialize_field("data", r)?;
                 t.end()
             }
+            ParagraphChild::Delete(ref r) => {
+                let mut t = serializer.serialize_struct("Delete", 2)?;
+                t.serialize_field("type", "delete")?;
+                t.serialize_field("data", r)?;
+                t.end()
+            }
+            ParagraphChild::BookmarkStart(ref r) => {
+                let mut t = serializer.serialize_struct("BookmarkStart", 2)?;
+                t.serialize_field("type", "bookmarkStart")?;
+                t.serialize_field("data", r)?;
+                t.end()
+            }
+            ParagraphChild::BookmarkEnd(ref r) => {
+                let mut t = serializer.serialize_struct("BookmarkEnd", 2)?;
+                t.serialize_field("type", "bookmarkEnd")?;
+                t.serialize_field("data", r)?;
+                t.end()
+            }
+            // TODO: Add comment later
             _ => {
                 let mut t = serializer.serialize_struct("Unsupported", 2)?;
                 t.serialize_field("type", "unsupported")?;

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)
 }
 

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);
     }
 

docx-core/src/reader/paragraph.rs

@@ -122,8 +122,7 @@ mod tests {
 
     #[test]
     fn test_read_indent() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
     <w:p>
         <w:pPr>
             <w:ind w:left="1470" w:right="1270" w:hanging="0"/>
@@ -161,8 +160,7 @@ mod tests {
 
     #[test]
     fn test_read_jc() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
     <w:p>
         <w:pPr>
             <w:jc w:val="left"/>
@@ -191,8 +189,7 @@ mod tests {
 
     #[test]
     fn test_read_numbering() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
     <w:p>
         <w:pPr>
             <w:numPr>
@@ -227,8 +224,7 @@ mod tests {
 
     #[test]
     fn test_read_insert() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
     <w:p>
         <w:ins w:id="0" w:author="unknown" w:date="2019-11-15T14:19:04Z">
             <w:r>
@@ -264,8 +260,7 @@ mod tests {
 
     #[test]
     fn test_read_delete() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
     <w:p>
         <w:del w:id="3" w:author="unknown" w:date="2019-11-15T14:19:04Z">
             <w:r>
@@ -301,8 +296,7 @@ mod tests {
 
     #[test]
     fn test_read_bookmark() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
     <w:p>
         <w:bookmarkStart w:id="0" w:name="ABCD-1234"/>
             <w:r>

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)
+}

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>

docx-core/src/reader/table.rs

@@ -72,8 +72,7 @@ mod tests {
 
     #[test]
     fn test_read_table_with_width_prop() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
 <w:tbl>
     <w:tblPr>
         <w:tblW w:w="9638" w:type="dxa"/>
@@ -97,8 +96,7 @@ mod tests {
 
     #[test]
     fn test_read_table_with_layout() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
 <w:tbl>
     <w:tblPr>
         <w:jc w:val="center"/>

docx-core/src/reader/table_cell.rs

@@ -94,8 +94,7 @@ mod tests {
 
     #[test]
     fn test_read_cell_with_prop() {
-        let c =
-            r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+        let c = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
 <w:tc>
     <w:tcPr>
         <w:tcW w:w="6425" w:type="dxa"/>

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),
         }
     }
 }

docx-core/src/types/break_type.rs

@@ -16,6 +16,7 @@ pub enum BreakType {
     Page,
     Column,
     TextWrapping,
+    Unsupported,
 }
 
 impl fmt::Display for BreakType {
@@ -24,6 +25,7 @@ impl fmt::Display for BreakType {
             BreakType::Page => write!(f, "page"),
             BreakType::Column => write!(f, "column"),
             BreakType::TextWrapping => write!(f, "textWrapping"),
+            BreakType::Unsupported => write!(f, "unsupported"),
         }
     }
 }
@@ -35,7 +37,7 @@ impl FromStr for BreakType {
             "page" => Ok(BreakType::Page),
             "column" => Ok(BreakType::Column),
             "textWrapping" => Ok(BreakType::TextWrapping),
-            _ => Err(errors::TypeError::FromStrError),
+            _ => Ok(BreakType::Unsupported),
         }
     }
 }

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),
         }
     }

docx-core/src/types/vertical_merge_type.rs

@@ -9,13 +9,15 @@ use std::str::FromStr;
 pub enum VMergeType {
     Continue,
     Restart,
+    Unsupported,
 }
 
 impl fmt::Display for VMergeType {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         match *self {
             VMergeType::Continue => write!(f, "continue"),
             VMergeType::Restart => write!(f, "restart"),
+            VMergeType::Unsupported => write!(f, "unsupported"),
         }
     }
 }
@@ -26,7 +28,7 @@ impl FromStr for VMergeType {
         match s {
             "continue" => Ok(VMergeType::Continue),
             "restart" => Ok(VMergeType::Restart),
-            _ => Err(errors::TypeError::FromStrError),
+            _ => Ok(VMergeType::Unsupported),
         }
     }
 }

docx-core/src/types/width_type.rs

@@ -11,6 +11,7 @@ pub enum WidthType {
     DXA,
     Auto,
     Pct,
+    Unsupported,
 }
 
 impl fmt::Display for WidthType {
@@ -19,6 +20,7 @@ impl fmt::Display for WidthType {
             WidthType::DXA => write!(f, "dxa"),
             WidthType::Auto => write!(f, "auto"),
             WidthType::Pct => write!(f, "pct"),
+            WidthType::Unsupported => write!(f, "unsupported"),
         }
     }
 }
@@ -30,7 +32,7 @@ impl FromStr for WidthType {
             "dxa" => Ok(WidthType::DXA),
             "auto" => Ok(WidthType::Auto),
             "pct" => Ok(WidthType::Pct),
-            _ => Err(errors::TypeError::FromStrError),
+            _ => Ok(WidthType::Unsupported),
         }
     }
 }