Switch to Poppler for PDF handling

assets/de.leopoldluley.Clapgrep.metainfo.xml.in.in

@@ -62,6 +62,20 @@
     <control>touch</control>
   </supports>
   <releases>
+    <release version="25.03" date="2025-03-01">
+      <description translate="no">
+        <p>New features:</p>
+        <ul>
+          <li>Implmented PDF preview.</li>
+          <li>Much faster PDF search using Poppler.</li>
+          <li>PDF results are now per line rather than per page.</li>
+        </ul>
+        <p>Bug fixes:</p>
+        <ul>
+          <li>None yet.</li>
+        </ul>
+      </description>
+    </release>
     <release version="25.02" date="2025-02-20">
       <description translate="no">
         <p>New features:</p>

build-aux/de.leopoldluley.Clapgrep.Devel.json

@@ -31,6 +31,52 @@
     "append-path": "/usr/lib/sdk/rust-stable/bin"
   },
   "modules": [
+    {
+      "name": "popplerdata",
+      "buildsystem": "cmake-ninja",
+      "sources": [
+        {
+          "type": "archive",
+          "url": "https://poppler.freedesktop.org/poppler-data-0.4.12.tar.gz",
+          "sha256": "c835b640a40ce357e1b83666aabd95edffa24ddddd49b8daff63adb851cdab74",
+          "x-checker-data": {
+            "type": "anitya",
+            "project-id": 3687,
+            "url-template": "https://poppler.freedesktop.org/poppler-data-$version.tar.gz"
+          }
+        }
+      ]
+    },
+    {
+      "name": "poppler",
+      "buildsystem": "cmake-ninja",
+      "config-opts": [
+        "-DCMAKE_INSTALL_LIBDIR=/app/lib",
+        "-DBUILD_GTK_TESTS=OFF",
+        "-DBUILD_CPP_TESTS=OFF",
+        "-DENABLE_CPP=OFF",
+        "-DENABLE_BOOST=OFF",
+        "-DENABLE_GOBJECT_INTROSPECTION=ON",
+        "-DENABLE_LIBOPENJPEG=openjpeg2",
+        "-DENABLE_QT5=OFF",
+        "-DENABLE_QT6=OFF"
+      ],
+      "cleanup": [
+        "/bin"
+      ],
+      "sources": [
+        {
+          "type": "archive",
+          "url": "https://poppler.freedesktop.org/poppler-25.02.0.tar.xz",
+          "sha256": "21234cb2a9647d73c752ce4031e65a79d11a511a835f2798284c2497b8701dee",
+          "x-checker-data": {
+            "type": "anitya",
+            "project-id": 3686,
+            "url-template": "https://poppler.freedesktop.org/poppler-$version.tar.xz"
+          }
+        }
+      ]
+    },
     {
       "name": "blueprint-compiler",
       "buildsystem": "meson",

core/Cargo.toml

@@ -13,5 +13,5 @@ flume = "0.11.1"
 dotext = "0.1.1"
 
 # for pdf support
-pdf-extract = "0.8.0"
-euclid = "0.20.14"     # must match pdf-extract::euclid version
+poppler-rs = "0.24.1"
+gio = "0.20.9"

core/src/extra/pdf.rs

@@ -1,11 +1,7 @@
 use crate::search::SearchSink;
-use euclid::vec2;
 use grep::{regex::RegexMatcher, searcher::Searcher};
-use pdf_extract::{
-    encryption::DecryptionError, ConvertToFmt, Document, MediaBox, OutputDev, OutputError,
-    Transform,
-};
-use std::{error::Error, fmt::Write, panic::catch_unwind, path::Path};
+use poppler::Document;
+use std::{error::Error, path::Path};
 
 pub static EXTENSIONS: &[&str] = &["pdf"];
 
@@ -15,122 +11,13 @@ pub fn process(
     path: &Path,
     sink: &mut SearchSink,
 ) -> Result<(), Box<dyn Error>> {
-    let text = extract(path)?;
-    searcher.search_slice(matcher, text.as_bytes(), sink)?;
-    Ok(())
-}
-
-fn extract(path: &Path) -> Result<String, Box<dyn Error>> {
-    let path = path.to_owned();
-    //because the library panics, we need to catch panics
-    let res = catch_unwind(|| extract_text(&path));
-    Ok(res.map_err(|_| "Panicked".to_string())??)
-}
-
-fn extract_text(path: impl AsRef<Path>) -> Result<String, OutputError> {
-    let mut s = String::new();
-    {
-        let mut output = PlainTextOutput::new(&mut s);
-        let doc = Document::load(path)?;
-        if doc.is_encrypted() {
-            return Err(OutputError::PdfError(pdf_extract::Error::Decryption(
-                DecryptionError::IncorrectPassword,
-            )));
+    let doc = Document::from_gfile(&gio::File::for_path(path), None, gio::Cancellable::NONE)?;
+    for i in 0..doc.n_pages() {
+        let page = doc.page(i).expect("out of range");
+        if let Some(text) = page.text() {
+            sink.page = Some(i as u64 + 1);
+            searcher.search_slice(matcher, text.as_bytes(), &mut *sink)?
         }
-        pdf_extract::output_doc(&doc, &mut output)?;
-    }
-    Ok(s)
-}
-
-struct PlainTextOutput<W: ConvertToFmt> {
-    writer: W::Writer,
-    last_end: f64,
-    last_y: f64,
-    first_char: bool,
-    flip_ctm: Transform,
-}
-
-impl<W: ConvertToFmt> PlainTextOutput<W> {
-    pub fn new(writer: W) -> PlainTextOutput<W> {
-        PlainTextOutput {
-            writer: writer.convert(),
-            last_end: 100000.,
-            first_char: false,
-            last_y: 0.,
-            flip_ctm: Transform::identity(),
-        }
-    }
-}
-
-type ArtBox = (f64, f64, f64, f64);
-
-/* There are some structural hints that PDFs can use to signal word and line endings:
- * however relying on these is not likely to be sufficient. */
-impl<W: ConvertToFmt> OutputDev for PlainTextOutput<W> {
-    fn begin_page(
-        &mut self,
-        _page_num: u32,
-        media_box: &MediaBox,
-        _: Option<ArtBox>,
-    ) -> Result<(), OutputError> {
-        self.flip_ctm = Transform::row_major(1., 0., 0., -1., 0., media_box.ury - media_box.lly);
-        Ok(())
-    }
-
-    fn end_page(&mut self) -> Result<(), OutputError> {
-        writeln!(self.writer)?;
-        Ok(())
-    }
-
-    fn output_character(
-        &mut self,
-        trm: &Transform,
-        width: f64,
-        _spacing: f64,
-        font_size: f64,
-        char: &str,
-    ) -> Result<(), OutputError> {
-        let position = trm.post_transform(&self.flip_ctm);
-        let transformed_font_size_vec = trm.transform_vector(vec2(font_size, font_size));
-        // get the length of one sized of the square with the same area with a rectangle of size (x, y)
-        let transformed_font_size =
-            (transformed_font_size_vec.x * transformed_font_size_vec.y).sqrt();
-        let (x, y) = (position.m31, position.m32);
-        if self.first_char {
-            if (y - self.last_y).abs() > transformed_font_size * 1.5 {
-                // writeln!(self.writer)?;
-                write!(self.writer, " ")?;
-            }
-
-            // we've moved to the left and down
-            if x < self.last_end && (y - self.last_y).abs() > transformed_font_size * 0.5 {
-                // writeln!(self.writer)?;
-                write!(self.writer, " ")?;
-            }
-
-            if x > self.last_end + transformed_font_size * 0.1 {
-                write!(self.writer, " ")?;
-            }
-        }
-        //let norm = unicode_normalization::UnicodeNormalization::nfkc(char);
-        write!(self.writer, "{}", char)?;
-        self.first_char = false;
-        self.last_y = y;
-        self.last_end = x + width * transformed_font_size;
-        Ok(())
-    }
-
-    fn begin_word(&mut self) -> Result<(), OutputError> {
-        self.first_char = true;
-        Ok(())
-    }
-
-    fn end_word(&mut self) -> Result<(), OutputError> {
-        Ok(())
-    }
-
-    fn end_line(&mut self) -> Result<(), OutputError> {
-        //write!(self.file, "\n");
-        Ok(())
     }
+    Ok(())
 }

core/src/search.rs

@@ -153,19 +153,22 @@ pub fn run(engine: SearchEngine, params: SearchParameters) {
 }
 
 pub struct SearchSink {
+    pub page: Option<u64>,
     matcher: RegexMatcher,
     entries: Vec<ResultEntry>,
 }
 
 impl SearchSink {
     pub fn new(matcher: RegexMatcher) -> Self {
         SearchSink {
+            page: None,
             matcher,
             entries: Vec::new(),
         }
     }
 
     pub fn take_entries(&mut self) -> Vec<ResultEntry> {
+        self.page = None;
         std::mem::take(&mut self.entries)
     }
 
@@ -199,10 +202,15 @@ impl grep::searcher::Sink for SearchSink {
         let content = String::from_utf8_lossy(mat.bytes())
             .trim_ascii_end()
             .to_string();
+
         let line = mat.line_number().unwrap();
+        let location = match self.page {
+            None => Location::Text { line },
+            Some(page) => Location::Document { page, line },
+        };
 
         self.entries.push(ResultEntry {
-            location: Location::Text { line },
+            location,
             content,
             matches,
         });

gnome/Cargo.toml

@@ -7,6 +7,7 @@ edition = "2021"
 clapgrep-core = { path = "../core" }
 adw = { version = "0.7.1", package = "libadwaita", features = ["v1_6"] }
 gtk = { version = "0.9.2", package = "gtk4", features = ["blueprint", "v4_16"] }
+poppler-rs = "0.24.1"
 flume = "0.11.1"
 gettext-rs = { version = "0.7.2", features = ["gettext-system"] }
 sourceview5 = { version = "0.9.1", features = ["gtk_v4_12", "v5_12"] }

gnome/src/search/result.rs

@@ -37,7 +37,7 @@ impl SearchResult {
 
         let line = match line {
             Location::Text { line } => line,
-            Location::Document { page: _, line } => line,
+            Location::Document { page, line: _ } => page,
         };
 
         let uri = if cfg!(target_os = "windows") {