Merge pull request #10 from randolf-scholz/main

Added .pre-commit-hooks.yaml, fixes #8 and #9

.pre-commit-hooks.yaml

@@ -0,0 +1,6 @@
+- id: nbstripout-fast
+  name: nbstripout-fast
+  entry: nbstripout-fast
+  types: [jupyter]
+  language: rust
+  description: "Strip output from Jupyter notebooks (modifies the files in place by default)."

examples/comparison.py

@@ -1,6 +1,5 @@
 #!/usr/bin/env python
 
-import json
 import re
 import subprocess
 import tempfile
@@ -32,27 +31,32 @@ def run(nb):
     return nb
 
 
-with tempfile.TemporaryDirectory() as tmpdir:
-    tmpdir = Path(tmpdir)
-    count_plus_filenames = []
-    for num_cells in [1, 10, 100, 1_000, 10_000]:
-        nb = run(create(num_cells))
-        filename = tmpdir / f"generated-{num_cells}-cells.ipynb"
-        count_plus_filenames.append((num_cells, filename))
-        with open(filename, "w") as f:
-            nbformat.write(nb, f)
-
-    print("{:<7} {:<12} {:<12}".format("Cells", "nbstripout", "nbstripout_fast"))
-    for num_cells, file in count_plus_filenames:
-        times = []
-        for cmd in [NBSTRIPOUT, NBSTRIPOUT_FAST]:
-            # emulate git filter by outputting to stdout
-            output = subprocess.check_output(
-                f"time {cmd} {file} -t > /dev/null",
-                stderr=subprocess.STDOUT,
-                universal_newlines=True,
-                shell=True,
-            )
-            real_time = re.match(r"real\s+(\w+\.\w+s)", output.strip()).group(1)
-            times.append(real_time)
-        print("{:<7} {:<12} {:<12}".format(num_cells, times[0], times[1]))
+def main():
+    with tempfile.TemporaryDirectory() as tmpdir:
+        tmpdir = Path(tmpdir)
+        count_plus_filenames = []
+        for num_cells in [1, 10, 100, 1_000, 10_000]:
+            nb = run(create(num_cells))
+            filename = tmpdir / f"generated-{num_cells}-cells.ipynb"
+            count_plus_filenames.append((num_cells, filename))
+            with open(filename, "w") as f:
+                nbformat.write(nb, f)
+
+        print("{:<7} {:<12} {:<12}".format("Cells", "nbstripout", "nbstripout_fast"))
+        for num_cells, file in count_plus_filenames:
+            times = []
+            for cmd in [NBSTRIPOUT, NBSTRIPOUT_FAST]:
+                # emulate git filter by outputting to stdout
+                output = subprocess.check_output(
+                    f"time {cmd} {file} -t > /dev/null",
+                    stderr=subprocess.STDOUT,
+                    universal_newlines=True,
+                    shell=True,
+                )
+                real_time = re.match(r"real\s+(\w+\.\w+s)", output.strip()).group(1)
+                times.append(real_time)
+            print("{:<7} {:<12} {:<12}".format(num_cells, times[0], times[1]))
+
+
+if __name__ == "__main__":
+    main()

src/main.rs

@@ -16,6 +16,7 @@ use std::env;
 use std::fs;
 use std::io;
 use std::io::BufRead;
+use std::path::PathBuf;
 
 mod stripoutlib;
 
@@ -58,7 +59,7 @@ struct Cli {
     keep_output: bool,
 
     #[clap(long, action)]
-    /// Remove cells where `source` is empty or contains only whitepace
+    /// Remove cells where `source` is empty or contains only whitespace
     drop_empty_cells: bool,
 
     #[clap(short, long, action)]
@@ -79,7 +80,7 @@ struct Cli {
 
     #[clap(parse(from_os_str))]
     /// Files to strip output from
-    files: Vec<std::path::PathBuf>,
+    files: Vec<PathBuf>,
 }
 
 #[derive(Deserialize, Debug)]
@@ -135,7 +136,7 @@ fn process_file(
     keep_count: bool,
     extra_keys: &Vec<String>,
     drop_empty_cells: bool,
-    output_file: Option<std::path::PathBuf>,
+    output_file: Option<PathBuf>,
 ) -> Result<(), String> {
     let mut nb: serde_json::Value = serde_json::from_str(&contents)
         .map_err(|e| format!("JSON was not well-formatted: {:?}", e))?;
@@ -155,18 +156,28 @@ fn process_file(
     let mut ser = serde_json::Serializer::with_formatter(buf, formatter);
     nb.serialize(&mut ser).map_err(|e| {
         format!(
-            "Unable to serialize notebook. Likely an intenral error: {:?}",
+            "Unable to serialize notebook. Likely an internal error: {:?}",
             e
         )
     })?;
-    let cleaned_contents = String::from_utf8(ser.into_inner()).map_err(|e| format!("{:?}", e))?;
+    let mut cleaned_contents = String::from_utf8(ser.into_inner()).map_err(|e| format!("{:?}", e))?;
 
-    if let Some(file) = output_file {
-        fs::write(&file, cleaned_contents)
-            .map_err(|e| format!("Could not write to {:?} due to {:?}", file, e))?;
+    // Check if the original content ended with a newline and the cleaned content doesn't
+    if contents.ends_with('\n') && !cleaned_contents.ends_with('\n') {
+        cleaned_contents.push('\n'); // Append a newline if necessary
+    }
+
+    if cleaned_contents != *contents {
+        if let Some(file) = output_file {
+            fs::write(&file, cleaned_contents)
+                .map_err(|e| format!("Could not write to {:?} due to {:?}", file, e))?;
+        } else {
+            println!("{}", cleaned_contents);
+        }
     } else {
-        println!("{}", cleaned_contents);
+        log::debug!("Content unchanged. File not modified.");
     }
+
     Ok(())
 }
 
@@ -197,7 +208,7 @@ fn main() -> Result<(), String> {
             }
             if let Some(config_keep_keys) = nbstripout_fast.keep_keys {
                 for key in config_keep_keys {
-                    // Remove all occurances
+                    // Remove all occurrences
                     extra_keys.retain(|x| x != &key);
                 }
             }