Parse via namespace and prefix

Documents are free to define a local prefix for a given namespace so parsing should not hard-code a specific prefix (e.g. "dc" for Dublin Core). This commit records all configured namespaces during the initial parse and uses these namespaces to instantiate the appropriate extensions as they are encountered.
rust-syndication · Apr 27, 2019 · 91c0c03 · 91c0c03
1 parent 7f4c387
commit 91c0c03
Show file tree

Hide file tree

Showing 6 changed files with 129 additions and 74 deletions.
diff --git a/src/channel.rs b/src/channel.rs
@@ -955,15 +955,11 @@ impl Channel {
                     if element.name() == b"rss" || element.name() == b"rdf:RDF" {
                         for attr in element.attributes().with_checks(false) {
                             if let Ok(attr) = attr {
-                                if !attr.key.starts_with(b"xmlns:") || attr.key == b"xmlns:itunes"
-                                    || attr.key == b"xmlns:dc"
-                                {
-                                    continue;
+                                if attr.key.starts_with(b"xmlns:") {
+                                    let prefix = str::from_utf8(&attr.key[6..])?.to_string();
+                                    let namespace = attr.unescape_and_decode_value(&reader)?;
+                                    namespaces.insert(prefix, namespace);
                                 }
-
-                                let key = str::from_utf8(&attr.key[6..])?.to_string();
-                                let value = attr.unescape_and_decode_value(&reader)?;
-                                namespaces.insert(key, value);
                             }
                         }
 
@@ -981,11 +977,11 @@ impl Channel {
             match reader.read_event(&mut buf)? {
                 Event::Start(element) => match element.name() {
                     b"channel" => {
-                        let inner = Channel::from_xml(&mut reader, element.attributes())?;
+                        let inner = Channel::from_xml(&namespaces, &mut reader, element.attributes())?;
                         channel = Some(inner);
                     }
                     b"item" => {
-                        let item = Item::from_xml(&mut reader, element.attributes())?;
+                        let item = Item::from_xml(&namespaces, &mut reader, element.attributes())?;
                         if items.is_none() {
                             items = Some(Vec::new());
                         }
@@ -1117,7 +1113,7 @@ impl ToString for Channel {
 }
 
 impl Channel {
-    pub fn from_xml<R: BufRead>(reader: &mut Reader<R>, _: Attributes) -> Result<Self, Error> {
+    pub fn from_xml<R: BufRead>(namespaces: &HashMap<String, String>, reader: &mut Reader<R>, _: Attributes) -> Result<Self, Error> {
         let mut channel = Channel::default();
         let mut buf = Vec::new();
         let mut skip_buf = Vec::new();
@@ -1142,7 +1138,7 @@ impl Channel {
                         channel.text_input = Some(text_input);
                     }
                     b"item" => {
-                        let item = Item::from_xml(reader, element.attributes())?;
+                        let item = Item::from_xml(&namespaces, reader, element.attributes())?;
                         channel.items.push(item);
                     }
                     b"title" => {
@@ -1229,12 +1225,17 @@ impl Channel {
         }
 
         if !channel.extensions.is_empty() {
-            if let Some(map) = channel.extensions.remove("itunes") {
-                channel.itunes_ext = Some(ITunesChannelExtension::from_map(map)?);
-            }
-
-            if let Some(map) = channel.extensions.remove("dc") {
-                channel.dublin_core_ext = Some(DublinCoreExtension::from_map(map));
+            // Process each of the namespaces we know (note that the values are not removed prior and reused to support pass-through of unknown extensions)
+            for (prefix, namespace) in namespaces {
+                match namespace.as_ref() {
+                    "http://www.itunes.com/dtds/podcast-1.0.dtd" => {
+                        channel.extensions.remove(prefix).map(|v| channel.itunes_ext = Some(ITunesChannelExtension::from_map(v)))
+                    },
+                    "http://purl.org/dc/elements/1.1/" => {
+                        channel.extensions.remove(prefix).map(|v| channel.dublin_core_ext = Some(DublinCoreExtension::from_map(v)))
+                    },
+                    _ => None
+                };
             }
         }
 

diff --git a/src/extension/itunes/itunes_channel_extension.rs b/src/extension/itunes/itunes_channel_extension.rs
@@ -13,7 +13,6 @@ use quick_xml::events::{BytesStart, Event};
 use quick_xml::Writer;
 
 use super::{parse_categories, parse_image, parse_owner};
-use error::Error;
 use extension::Extension;
 use extension::itunes::{ITunesCategory, ITunesOwner};
 use extension::util::remove_extension_value;
@@ -436,20 +435,20 @@ impl ITunesChannelExtension {
 
 impl ITunesChannelExtension {
     /// Create an `ITunesChannelExtension` from a `HashMap`.
-    pub fn from_map(mut map: HashMap<String, Vec<Extension>>) -> Result<Self, Error> {
+    pub fn from_map(mut map: HashMap<String, Vec<Extension>>) -> Self {
         let mut ext = ITunesChannelExtension::default();
         ext.author = remove_extension_value(&mut map, "author");
         ext.block = remove_extension_value(&mut map, "block");
-        ext.categories = parse_categories(&mut map)?;
+        ext.categories = parse_categories(&mut map);
         ext.image = parse_image(&mut map);
         ext.explicit = remove_extension_value(&mut map, "explicit");
         ext.complete = remove_extension_value(&mut map, "complete");
         ext.new_feed_url = remove_extension_value(&mut map, "new-feed-url");
-        ext.owner = parse_owner(&mut map)?;
+        ext.owner = parse_owner(&mut map);
         ext.subtitle = remove_extension_value(&mut map, "subtitle");
         ext.summary = remove_extension_value(&mut map, "summary");
         ext.keywords = remove_extension_value(&mut map, "keywords");
-        Ok(ext)
+        ext
     }
 }
 

diff --git a/src/extension/itunes/mod.rs b/src/extension/itunes/mod.rs
@@ -7,7 +7,6 @@
 
 use std::collections::HashMap;
 
-use error::Error;
 use extension::Extension;
 
 mod itunes_category;
@@ -32,12 +31,10 @@ fn parse_image(map: &mut HashMap<String, Vec<Extension>>) -> Option<String> {
     element.attrs.remove("href")
 }
 
-fn parse_categories(
-    map: &mut HashMap<String, Vec<Extension>>,
-) -> Result<Vec<ITunesCategory>, Error> {
+fn parse_categories(map: &mut HashMap<String, Vec<Extension>>) -> Vec<ITunesCategory> {
     let mut elements = match map.remove("category") {
         Some(elements) => elements,
-        None => return Ok(Vec::new()),
+        None => return Vec::new(),
     };
 
     let mut categories = Vec::with_capacity(elements.len());
@@ -62,27 +59,26 @@ fn parse_categories(
         categories.push(category);
     }
 
-    Ok(categories)
+    categories
 }
 
-fn parse_owner(map: &mut HashMap<String, Vec<Extension>>) -> Result<Option<ITunesOwner>, Error> {
-    let mut element = match map.remove("owner").map(|mut v| v.remove(0)) {
-        Some(element) => element,
-        None => return Ok(None),
-    };
-
-    let name = element
-        .children
-        .remove("name")
-        .and_then(|mut v| v.remove(0).value);
-
-    let email = element
-        .children
-        .remove("email")
-        .and_then(|mut v| v.remove(0).value);
-
-    let mut owner = ITunesOwner::default();
-    owner.set_name(name);
-    owner.set_email(email);
-    Ok(Some(owner))
+fn parse_owner(map: &mut HashMap<String, Vec<Extension>>) -> Option<ITunesOwner> {
+    if let Some(mut element) = map.remove("owner").map(|mut v| v.remove(0)) {
+        let name = element
+            .children
+            .remove("name")
+            .and_then(|mut v| v.remove(0).value);
+
+        let email = element
+            .children
+            .remove("email")
+            .and_then(|mut v| v.remove(0).value);
+
+        let mut owner = ITunesOwner::default();
+        owner.set_name(name);
+        owner.set_email(email);
+        Some(owner)
+    } else {
+        None
+    }
 }
diff --git a/src/item.rs b/src/item.rs
@@ -24,6 +24,7 @@ use guid::Guid;
 use source::Source;
 use toxml::{ToXml, WriterExt};
 use util::element_text;
+use std::collections::HashMap;
 
 /// Represents an item in an RSS feed.
 #[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
@@ -536,7 +537,7 @@ impl Item {
 }
 
 impl Item {
-    pub fn from_xml<R: BufRead>(reader: &mut Reader<R>, _: Attributes) -> Result<Self, Error> {
+    pub fn from_xml<R: BufRead>(namespaces: &HashMap<String, String>, reader: &mut Reader<R>, _: Attributes) -> Result<Self, Error> {
         let mut item = Item::default();
         let mut buf = Vec::new();
 
@@ -588,12 +589,17 @@ impl Item {
         }
 
         if !item.extensions.is_empty() {
-            if let Some(map) = item.extensions.remove("itunes") {
-                item.itunes_ext = Some(ITunesItemExtension::from_map(map));
-            }
-
-            if let Some(map) = item.extensions.remove("dc") {
-                item.dublin_core_ext = Some(DublinCoreExtension::from_map(map));
+            // Process each of the namespaces we know (note that the values are not removed prior and reused to support pass-through of unknown extensions)
+            for (prefix, namespace) in namespaces {
+                match namespace.as_ref() {
+                    "http://www.itunes.com/dtds/podcast-1.0.dtd" => {
+                        item.extensions.remove(prefix).map(|v| item.itunes_ext = Some(ITunesItemExtension::from_map(v)))
+                    },
+                    "http://purl.org/dc/elements/1.1/" => {
+                        item.extensions.remove(prefix).map(|v| item.dublin_core_ext = Some(DublinCoreExtension::from_map(v)))
+                    },
+                    _ => None
+                };
             }
         }
 

diff --git a/tests/data/dublincore_altprefix.xml b/tests/data/dublincore_altprefix.xml
@@ -0,0 +1,39 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dublincore="http://purl.org/dc/elements/1.1/">
+	<channel>
+		<dublincore:contributor>Contributor 1</dublincore:contributor>
+		<dublincore:contributor>Contributor 2</dublincore:contributor>
+		<dublincore:coverage>Coverage</dublincore:coverage>
+		<dublincore:creator>Creator</dublincore:creator>
+		<dublincore:date>2016-08-27</dublincore:date>
+		<dublincore:description>Description</dublincore:description>
+		<dublincore:format>text/plain</dublincore:format>
+		<dublincore:identifier>Identifier</dublincore:identifier>
+		<dublincore:language>en-US</dublincore:language>
+		<dublincore:publisher>Publisher</dublincore:publisher>
+		<dublincore:relation>Relation</dublincore:relation>
+		<dublincore:rights>Company</dublincore:rights>
+		<dublincore:source>Source</dublincore:source>
+		<dublincore:subject>Subject</dublincore:subject>
+		<dublincore:title>Title</dublincore:title>
+		<dublincore:type>Type</dublincore:type>
+		<item>
+			<dublincore:contributor>Contributor 1</dublincore:contributor>
+			<dublincore:contributor>Contributor 2</dublincore:contributor>
+			<dublincore:coverage>Coverage</dublincore:coverage>
+			<dublincore:creator>Creator</dublincore:creator>
+			<dublincore:date>2016-08-27</dublincore:date>
+			<dublincore:description>Description</dublincore:description>
+			<dublincore:format>text/plain</dublincore:format>
+			<dublincore:identifier>Identifier</dublincore:identifier>
+			<dublincore:language>en-US</dublincore:language>
+			<dublincore:publisher>Publisher</dublincore:publisher>
+			<dublincore:relation>Relation</dublincore:relation>
+			<dublincore:rights>Company</dublincore:rights>
+			<dublincore:source>Source</dublincore:source>
+			<dublincore:subject>Subject</dublincore:subject>
+			<dublincore:title>Title</dublincore:title>
+			<dublincore:type>Type</dublincore:type>
+		</item>
+	</channel>
+</rss>