pdf: add page number prefix

src/adapters/poppler.rs

@@ -1,6 +1,7 @@
 use super::*;
 use lazy_static::lazy_static;
 use spawning::SpawningFileAdapter;
+use std::io::BufReader;
 use std::process::Command;
 
 static EXTENSIONS: &[&str] = &["pdf"];
@@ -29,6 +30,20 @@ impl GetMetadata for PopplerAdapter {
     }
 }
 impl SpawningFileAdapter for PopplerAdapter {
+    fn postproc(line_prefix: &str, inp: &mut Read, oup: &mut Write) -> Fallible<()> {
+        // prepend Page X to each line
+        let mut page = 1;
+        for line in BufReader::new(inp).lines() {
+            let mut line = line?;
+            if line.contains("\x0c") {
+                // page break
+                line = line.replace("\x0c", "");
+                page += 1;
+            }
+            oup.write_all(format!("{}Page {}: {}\n", line_prefix, page, line).as_bytes())?;
+        }
+        Ok(())
+    }
     fn get_exe(&self) -> &str {
         "pdftotext"
     }

src/adapters/zip.rs

@@ -53,7 +53,7 @@ impl FileAdapter for ZipAdapter {
                         file.size(),
                         file.compressed_size()
                     );
-                    let line_prefix = &format!("{}{}:/", line_prefix, file.name().clone());
+                    let line_prefix = &format!("{}{}: ", line_prefix, file.name().clone());
                     rga_preproc(
                         AdaptInfo {
                             filepath_hint: &file.sanitized_name(),

src/bin/rga-preproc.rs

@@ -1,11 +1,8 @@
 use failure::{format_err, Error};
-use path_clean::PathClean;
 use rga::adapters::*;
 use rga::preproc::*;
-use rga::CachingWriter;
+use std::env;
 use std::fs::File;
-use std::path::PathBuf;
-use std::rc::Rc;
 
 fn main() -> Result<(), Error> {
     let path = {
@@ -26,5 +23,11 @@ fn main() -> Result<(), Error> {
         line_prefix: "",
     };
 
-    rga_preproc(ai, Some(open_cache_db()?))
+    let cache_db = match env::var("RGA_NO_CACHE") {
+        Ok(ref s) if s.len() > 0 => Some(open_cache_db()?),
+        Ok(_) => None,
+        Err(_) => None,
+    };
+
+    rga_preproc(ai, cache_db)
 }

src/bin/rga.rs

@@ -14,14 +14,10 @@ fn main() -> std::io::Result<()> {
             "--list-adapters 'Lists all known adapters'",
         ))
         .arg(Arg::from_usage("--adapters=[commaseparated] 'Change which adapters to use and in which priority order (descending)'").require_equals(true))
+        .arg(Arg::from_usage("--no-cache 'Disable caching of results'"))
         .arg(Arg::from_usage("--rg-help 'Show help for ripgrep itself'"))
         .arg(Arg::from_usage("--rg-version 'Show version of ripgrep itself'"));
     app.p.create_help_and_version();
-    println!("g={:#?},f={:#?}", app.p.groups, app.p.flags);
-    for opt in app.p.opts() {
-        println!("opt {:#?}", opt.s.long);
-    }
-    //
     let mut firstarg = true;
     let (our_args, mut passthrough_args): (Vec<OsString>, Vec<OsString>) = std::env::args_os()
         .partition(|os_arg| {
@@ -71,7 +67,16 @@ fn main() -> std::io::Result<()> {
         println!("Adapters:");
         for adapter in adapters {
             let meta = adapter.metadata();
-            println!("{} v{}", meta.name, meta.version);
+            let matchers = meta
+                .matchers
+                .iter()
+                .map(|m| match m {
+                    Matcher::FileExtension(ext) => format!(".{}", ext),
+                })
+                .collect::<Vec<_>>()
+                .join(", ");
+            print!("{} v{}: {}", meta.name, meta.version, matchers);
+            println!("");
         }
         return Ok(());
     }