Merge pull request #60 from AtkinsSJ/read-lines

Extract code for reading a line at a time, from head/tail/wc builtins
HeyPuter · Mar 14, 2024 · 56be4f1 · 56be4f1
2 parents dd88f33 + 08f8724
commit 56be4f1
Show file tree

Hide file tree

Showing 4 changed files with 54 additions and 80 deletions.
diff --git a/src/puter-shell/coreutils/head.js b/src/puter-shell/coreutils/head.js
@@ -17,7 +17,7 @@
  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  */
 import { Exit } from './coreutil_lib/exit.js';
-import { resolveRelativePath } from '../../util/path.js';
+import { fileLines } from '../../util/file.js';
 
 export default {
     name: 'head',
@@ -41,9 +41,8 @@ export default {
         }
     },
     execute: async ctx => {
-        const { in_, out, err } = ctx.externs;
+        const { out, err } = ctx.externs;
         const { positionals, values } = ctx.locals;
-        const { filesystem } = ctx.platform;
 
         if (positionals.length > 1) {
             // TODO: Support multiple files (this is POSIX)
@@ -63,32 +62,12 @@ export default {
             lineCount = parsedLineCount;
         }
 
-        // TODO: head can stop reading from the input as soon as it completes lineCount lines.
-        let lines = [];
-        if (relPath === '-') {
-            lines = await in_.collect();
-        } else {
-            const absPath = resolveRelativePath(ctx.vars, relPath);
-            const fileData = await filesystem.read(absPath);
-            // DRY: Similar logic in wc and tail
-            if (fileData instanceof Blob) {
-                const arrayBuffer = await fileData.arrayBuffer();
-                const fileText = new TextDecoder().decode(arrayBuffer);
-                lines = fileText.split(/\n|\r|\r\n/).map(it => it + '\n');
-            } else if (typeof fileData === 'string') {
-                lines = fileData.split(/\n|\r|\r\n/).map(it => it + '\n');
-            } else {
-                // ArrayBuffer or TypedArray
-                const fileText = new TextDecoder().decode(fileData);
-                lines = fileText.split(/\n|\r|\r\n/).map(it => it + '\n');
-            }
-        }
-        if ( lines.length > lineCount ) {
-            lines = lines.slice(0, lineCount);
-        }
-
-        for ( const line of lines ) {
+        let processedLineCount = 0;
+        for await (const line of fileLines(ctx, relPath)) {
             await out.write(line);
+            processedLineCount++;
+            if (processedLineCount >= lineCount)
+                break;
         }
     }
 };
diff --git a/src/puter-shell/coreutils/tail.js b/src/puter-shell/coreutils/tail.js
@@ -17,7 +17,7 @@
  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  */
 import { Exit } from './coreutil_lib/exit.js';
-import { resolveRelativePath } from '../../util/path.js';
+import { fileLines } from '../../util/file.js';
 
 export default {
     name: 'tail',
@@ -41,9 +41,8 @@ export default {
         }
     },
     execute: async ctx => {
-        const { in_, out, err } = ctx.externs;
+        const { out, err } = ctx.externs;
         const { positionals, values } = ctx.locals;
-        const { filesystem } = ctx.platform;
 
         if (positionals.length > 1) {
             // TODO: Support multiple files (this is an extension to POSIX, but available in the GNU tail)
@@ -64,30 +63,21 @@ export default {
         }
 
         let lines = [];
-        if (relPath === '-') {
-            lines = await in_.collect();
-        } else {
-            const absPath = resolveRelativePath(ctx.vars, relPath);
-            const fileData = await filesystem.read(absPath);
-            // DRY: Similar logic in wc
-            if (fileData instanceof Blob) {
-                const arrayBuffer = await fileData.arrayBuffer();
-                const fileText = new TextDecoder().decode(arrayBuffer);
-                lines = fileText.split(/\n|\r|\r\n/).map(it => it + '\n');
-            } else if (typeof fileData === 'string') {
-                lines = fileData.split(/\n|\r|\r\n/).map(it => it + '\n');
-            } else {
-                // ArrayBuffer or TypedArray
-                const fileText = new TextDecoder().decode(fileData);
-                lines = fileText.split(/\n|\r|\r\n/).map(it => it + '\n');
+        for await (const line of fileLines(ctx, relPath)) {
+            lines.push(line);
+            // We keep lineCount+1 lines, to account for a possible trailing blank line.
+            if (lines.length > lineCount + 1) {
+                lines.shift();
             }
         }
+
+        // Ignore trailing blank line
         if ( lines.length > 0 && lines[lines.length - 1] === '\n') {
-            // Ignore trailing blank line
             lines.pop();
         }
+        // Now we remove the extra line if it's there.
         if ( lines.length > lineCount ) {
-            lines = lines.slice(-1 * lineCount);
+            lines.shift();
         }
 
         for ( const line of lines ) {

diff --git a/src/puter-shell/coreutils/wc.js b/src/puter-shell/coreutils/wc.js
@@ -17,6 +17,7 @@
  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  */
 import { resolveRelativePath } from '../../util/path.js';
+import { fileLines } from '../../util/file.js';
 
 const TAB_SIZE = 8;
 
@@ -100,26 +101,15 @@ export default {
 
             let inWord = false;
             let currentLineLength = 0;
-            let accumulateData = async (input) => {
-                let stringInput;
-                if (input instanceof Blob) {
-                    const arrayBuffer = await input.arrayBuffer();
-                    stringInput = new TextDecoder().decode(arrayBuffer);
-                    counts.bytes += arrayBuffer.byteLength;
-                } else if (typeof input === 'string') {
-                    stringInput = input;
-                    if (printBytes) {
-                        const byteInput = new TextEncoder().encode(input);
-                        counts.bytes += byteInput.length;
-                    }
-                } else {
-                    // ArrayBuffer or TypedArray
-                    stringInput = new TextDecoder().decode(input);
-                    counts.bytes += input.length;
+
+            for await (const line of fileLines(ctx, relPath)) {
+                counts.chars += line.length;
+                if (printBytes) {
+                    const byteInput = new TextEncoder().encode(line);
+                    counts.bytes += byteInput.length;
                 }
-                counts.chars += stringInput.length;
 
-                for (const char of stringInput) {
+                for (const char of line) {
                     // "The wc utility shall consider a word to be a non-zero-length string of characters delimited by white space."
                     if (/\s/.test(char)) {
                         if (char === '\r' || char === '\n') {
@@ -142,19 +132,6 @@ export default {
                 }
             }
 
-            if (relPath === '-') {
-                let chunk, done;
-                const nextChunk = async () => {
-                    ({ value: chunk, done } = await ctx.externs.in_.read());
-                }
-                for ( await nextChunk() ; ! done ; await nextChunk() ) {
-                    await accumulateData(chunk);
-                }
-            } else {
-                const absPath = resolveRelativePath(ctx.vars, relPath);
-                const fileData = await filesystem.read(absPath);
-                await accumulateData(fileData);
-            }
             counts.maxLineLength = Math.max(counts.maxLineLength, currentLineLength);
 
             newlinesWidth = Math.max(newlinesWidth, counts.newlines.toString().length);

diff --git a/src/util/file.js b/src/util/file.js
@@ -0,0 +1,28 @@
+import { resolveRelativePath } from './path.js';
+
+// Iterate the given file, one line at a time.
+// TODO: Make this read one line at a time, instead of all at once.
+export async function* fileLines(ctx, relPath, options = { dashIsStdin: true }) {
+    let lines = [];
+    if (options.dashIsStdin && relPath === '-') {
+        lines = await ctx.externs.in_.collect();
+    } else {
+        const absPath = resolveRelativePath(ctx.vars, relPath);
+        const fileData = await ctx.platform.filesystem.read(absPath);
+        if (fileData instanceof Blob) {
+            const arrayBuffer = await fileData.arrayBuffer();
+            const fileText = new TextDecoder().decode(arrayBuffer);
+            lines = fileText.split(/\n|\r|\r\n/).map(it => it + '\n');
+        } else if (typeof fileData === 'string') {
+            lines = fileData.split(/\n|\r|\r\n/).map(it => it + '\n');
+        } else {
+            // ArrayBuffer or TypedArray
+            const fileText = new TextDecoder().decode(fileData);
+            lines = fileText.split(/\n|\r|\r\n/).map(it => it + '\n');
+        }
+    }
+
+    for (const line of lines) {
+        yield line;
+    }
+}