@storyteller-platform/align 0.1.57 → 0.1.58

package/dist/cli/bin.cjs

@@ -69,27 +69,31 @@ var __callDispose = (stack, error, hasError) => {
 };
 var import_node_crypto = require("node:crypto");
 var import_node_fs = require("node:fs");
-var os = __toESM(require("node:os"), 1);
 var import_node_path = require("node:path");
 var import_core = require("@optique/core");
 var import_run = require("@optique/run");
 var import_valueparser = require("@optique/run/valueparser");
 var import_cli_progress = require("cli-progress");
 var import_epub = require("@storyteller-platform/epub");
+var import_fs = require("@storyteller-platform/fs");
 var import_package = __toESM(require("../../package.json"), 1);
 var import_align = require("../align/align.cjs");
+var import_align2 = require("../align/ctc/align.cjs");
 var import_parse = require("../align/parse.cjs");
 var import_logging = require("../common/logging.cjs");
 var import_parse2 = require("../common/parse.cjs");
+var import_emit = require("../emit/emit.cjs");
+var import_parse3 = require("../emit/parse.cjs");
 var import_markup = require("../markup/markup.cjs");
-var import_parse3 = require("../markup/parse.cjs");
-var import_parse4 = require("../process/parse.cjs");
+var import_parse4 = require("../markup/parse.cjs");
+var import_parse5 = require("../process/parse.cjs");
 var import_processAudiobook = require("../process/processAudiobook.cjs");
-var import_parse5 = require("../snapshot/parse.cjs");
-var import_snapshot = require("../snapshot/snapshot.cjs");
-var import_parse6 = require("../transcribe/parse.cjs");
+var import_snapshot = require("../snapshot/ctc/snapshot.cjs");
+var import_parse6 = require("../snapshot/parse.cjs");
+var import_snapshot2 = require("../snapshot/snapshot.cjs");
+var import_parse7 = require("../transcribe/parse.cjs");
 var import_transcribe = require("../transcribe/transcribe.cjs");
-var import_parse7 = require("../upgradeEpub/parse.cjs");
+var import_parse8 = require("../upgradeEpub/parse.cjs");
 var import_upgradeEpub = require("../upgradeEpub/upgradeEpub.cjs");
 const pipelineCommand = (0, import_core.merge)(
   (0, import_core.object)({
@@ -97,9 +101,6 @@ const pipelineCommand = (0, import_core.merge)(
     processedAudio: (0, import_core.optional)(
       (0, import_core.option)("--processed-audio", (0, import_valueparser.path)({ type: "directory" }))
     ),
-    transcriptions: (0, import_core.optional)(
-      (0, import_core.option)("--transcriptions", (0, import_valueparser.path)({ type: "directory" }))
-    ),
     markedup: (0, import_core.optional)(
       (0, import_core.option)("--markedup", (0, import_valueparser.path)({ type: "file", extensions: [".epub"] }))
     ),
@@ -113,8 +114,26 @@ const pipelineCommand = (0, import_core.merge)(
     ),
     output: (0, import_core.option)("--output", (0, import_valueparser.path)({ type: "file", extensions: [".epub"] }))
   }),
-  import_parse4.processParser,
-  (0, import_core.group)("Transcription", import_parse6.transcribeParser),
+  import_parse5.processParser,
+  (0, import_core.or)(
+    (0, import_core.merge)(
+      (0, import_core.object)({
+        ctc: (0, import_core.flag)("--ctc", {
+          description: import_core.message`Use CTC emissions instead of transcriptions for alignment`
+        }),
+        emissions: (0, import_core.optional)((0, import_core.option)("--emissions", (0, import_valueparser.path)({ type: "directory" })))
+      }),
+      import_parse3.emitParser
+    ),
+    (0, import_core.merge)(
+      (0, import_core.group)("Transcription", import_parse7.transcribeParser),
+      (0, import_core.object)({
+        transcriptions: (0, import_core.optional)(
+          (0, import_core.option)("--transcriptions", (0, import_valueparser.path)({ type: "directory" }))
+        )
+      })
+    )
+  ),
   import_parse2.autoUpgradeParser,
   import_parse2.removeNcxParser,
   import_parse2.granularityParser,
@@ -123,17 +142,25 @@ const pipelineCommand = (0, import_core.merge)(
   import_parse2.loggingParser
 );
 const parser = (0, import_core.or)(
-  import_parse4.processCommand,
-  import_parse6.transcribeCommand,
-  import_parse7.upgradeEpubCommand,
-  import_parse3.markupCommand,
+  import_parse5.processCommand,
+  import_parse7.transcribeCommand,
+  import_parse3.emitCommand,
+  import_parse8.upgradeEpubCommand,
+  import_parse4.markupCommand,
   import_parse.alignCommand,
   pipelineCommand,
-  import_parse5.snapshotCommand
+  import_parse6.snapshotCommand
 );
 async function main() {
   var _stack2 = [];
   try {
+    process.on("SIGINT", () => {
+      try {
+        progressBar.stop();
+      } finally {
+        process.exit(130);
+      }
+    });
     const parsed = (0, import_run.run)(parser, {
       showChoices: true,
       showDefault: true,
@@ -157,6 +184,17 @@ async function main() {
         progressBar.start(100, 0);
       }
     }
+    function updateProgressBar(progress) {
+      if (parsed.action === "upgrade-epub") return;
+      if (parsed.noProgress || parsed.logLevel !== "silent") {
+        return;
+      }
+      const floored = Math.floor(progress * 100);
+      if (floored === Math.floor(progressBar.getProgress() * 100)) {
+        return;
+      }
+      progressBar.update(floored);
+    }
     resetProgressBar();
     process.on("SIGINT", () => {
       controller.abort();
@@ -178,11 +216,7 @@ async function main() {
           parallelism: parsed.parallelism,
           signal: controller.signal,
           logger,
-          ...!parsed.noProgress && parsed.logLevel === "silent" && {
-            onProgress: (progress) => {
-              progressBar.update(Math.floor(progress * 100));
-            }
-          }
+          onProgress: updateProgressBar
         });
         if (parsed.time) {
           timing.print("Process audiobook");
@@ -198,11 +232,7 @@ async function main() {
             ...parsed,
             signal: controller.signal,
             logger,
-            ...!parsed.noProgress && parsed.logLevel === "silent" && {
-              onProgress: (progress) => {
-                progressBar.update(Math.floor(progress * 100));
-              }
-            }
+            onProgress: updateProgressBar
           }
         );
         if (parsed.time) {
@@ -210,6 +240,13 @@ async function main() {
         }
         break;
       }
+      case "emit": {
+        await (0, import_emit.emit)(parsed.input, parsed.output, {
+          ...parsed,
+          onProgress: updateProgressBar
+        });
+        break;
+      }
       case "upgrade-epub": {
         await (0, import_upgradeEpub.upgradeEpub)(parsed.input, parsed.output, {
           removeNcx: parsed.removeNcx
@@ -220,8 +257,7 @@ async function main() {
         let input = parsed.input;
         if (parsed.autoupgrade) {
           logger.info("Upgrading EPUB 2 to EPUB 3");
-          input = (0, import_node_path.join)(
-            os.tmpdir(),
+          input = (0, import_fs.scratchPath)(
             `stalign-autoupgrade-${(0, import_node_crypto.randomUUID)()}`,
             (0, import_node_path.basename)(parsed.input)
           );
@@ -231,11 +267,7 @@ async function main() {
           primaryLocale: parsed.language ?? new Intl.Locale("en-US"),
           granularity: parsed.granularity,
           logger,
-          ...!parsed.noProgress && parsed.logLevel === "silent" && {
-            onProgress: (progress) => {
-              progressBar.update(Math.floor(progress * 100));
-            }
-          }
+          onProgress: updateProgressBar
         });
         if (parsed.time) {
           timing.print("Mark up EPUB");
@@ -246,34 +278,50 @@ async function main() {
         let input = parsed.epub;
         if (parsed.autoupgrade) {
           logger.info("Upgrading EPUB 2 to EPUB 3");
-          input = (0, import_node_path.join)(
-            os.tmpdir(),
+          input = (0, import_fs.scratchPath)(
             `stalign-autoupgrade-${(0, import_node_crypto.randomUUID)()}`,
             (0, import_node_path.basename)(parsed.epub)
           );
           await (0, import_upgradeEpub.upgradeEpub)(parsed.epub, input, { removeNcx: parsed.removeNcx });
         }
-        const timing = await (0, import_align.align)(
-          input,
-          parsed.output,
-          parsed.transcriptions,
-          parsed.audiobook,
-          {
-            granularity: parsed.granularity,
-            textRef: parsed.textRef,
-            outFormat: parsed.outFormat,
-            primaryLocale: parsed.language,
-            reportsPath: parsed.reports,
-            logger,
-            ...!parsed.noProgress && parsed.logLevel === "silent" && {
-              onProgress: (progress) => {
-                progressBar.update(Math.floor(progress * 100));
-              }
+        if ("ctc" in parsed) {
+          const timing = await (0, import_align2.ctcAlign)(
+            input,
+            parsed.output,
+            parsed.emissions,
+            parsed.audiobook,
+            {
+              granularity: parsed.granularity,
+              textRef: parsed.textRef,
+              outFormat: parsed.outFormat,
+              primaryLocale: parsed.language,
+              reportsPath: parsed.reports,
+              logger,
+              onProgress: updateProgressBar
             }
+          );
+          if (parsed.time) {
+            timing.print("Align EPUB and audiobook");
+          }
+        } else {
+          const timing = await (0, import_align.align)(
+            input,
+            parsed.output,
+            parsed.transcriptions,
+            parsed.audiobook,
+            {
+              granularity: parsed.granularity,
+              textRef: parsed.textRef,
+              outFormat: parsed.outFormat,
+              primaryLocale: parsed.language,
+              reportsPath: parsed.reports,
+              logger,
+              onProgress: updateProgressBar
+            }
+          );
+          if (parsed.time) {
+            timing.print("Align EPUB and audiobook");
           }
-        );
-        if (parsed.time) {
-          timing.print("Align EPUB and audiobook");
         }
         break;
       }
@@ -283,8 +331,7 @@ async function main() {
           let input = parsed.epub;
           if (parsed.autoupgrade) {
             logger.info("Upgrading EPUB 2 to EPUB 3");
-            input = (0, import_node_path.join)(
-              os.tmpdir(),
+            input = (0, import_fs.scratchPath)(
               `stalign-autoupgrade-${(0, import_node_crypto.randomUUID)()}`,
               (0, import_node_path.basename)(parsed.epub)
             );
@@ -293,7 +340,7 @@ async function main() {
           const epub = __using(_stack, await import_epub.Epub.from(input));
           const primaryLocale = parsed.language ?? await epub.getLanguage() ?? new Intl.Locale("en-US");
           epub.discardAndClose();
-          const processedAudio = parsed.processedAudio ?? (0, import_node_path.join)(os.tmpdir(), `stalign-processed-${(0, import_node_crypto.randomUUID)()}`);
+          const processedAudio = parsed.processedAudio ?? (0, import_fs.scratchPath)(`stalign-processed-${(0, import_node_crypto.randomUUID)()}`);
           if (!parsed.processedAudio) {
             stack.defer(() => {
               (0, import_node_fs.rmSync)(processedAudio, { recursive: true, force: true });
@@ -310,11 +357,7 @@ async function main() {
               maxLength: parsed.maxLength / 60,
               parallelism: parsed.parallelTranscodes,
               signal: controller.signal,
-              ...!parsed.noProgress && parsed.logLevel === "silent" && {
-                onProgress: (progress) => {
-                  progressBar.update(Math.floor(progress * 100));
-                }
-              }
+              onProgress: updateProgressBar
             }
           );
           resetProgressBar();
@@ -326,36 +369,56 @@ async function main() {
           }
           logger.info("Transcribing...");
           startProgressBar();
-          const transcriptions = parsed.transcriptions ?? (0, import_node_path.join)(os.tmpdir(), `stalign-transcriptions-${(0, import_node_crypto.randomUUID)()}`);
-          if (!parsed.transcriptions) {
-            stack.defer(() => {
-              (0, import_node_fs.rmSync)(transcriptions, { recursive: true, force: true });
-            });
-          }
-          const transcribeTiming = await (0, import_transcribe.transcribe)(
-            processedAudio,
-            transcriptions,
-            primaryLocale,
-            {
+          let emissions;
+          let transcriptions;
+          if ("ctc" in parsed) {
+            emissions = parsed.emissions ?? (0, import_fs.scratchPath)(`stalign-emissions-${(0, import_node_crypto.randomUUID)()}`);
+            if (!parsed.emissions) {
+              stack.defer(() => {
+                (0, import_node_fs.rmSync)(emissions, { recursive: true, force: true });
+              });
+            }
+            const transcribeTiming = await (0, import_emit.emit)(processedAudio, emissions, {
               ...parsed,
-              parallelism: parsed.parallelTranscribes,
               signal: controller.signal,
               logger,
-              ...!parsed.noProgress && parsed.logLevel === "silent" && {
-                onProgress: (progress) => {
-                  progressBar.update(Math.floor(progress * 100));
-                }
+              onProgress: updateProgressBar
+            });
+            resetProgressBar();
+            logger.info(
+              `Generating emissions for audiobook complete, emissions saved to ${emissions}.`
+            );
+            if (parsed.time) {
+              transcribeTiming.print();
+            }
+          } else {
+            transcriptions = parsed.transcriptions ?? (0, import_fs.scratchPath)(`stalign-transcriptions-${(0, import_node_crypto.randomUUID)()}`);
+            if (!parsed.transcriptions) {
+              stack.defer(() => {
+                (0, import_node_fs.rmSync)(transcriptions, { recursive: true, force: true });
+              });
+            }
+            const transcribeTiming = await (0, import_transcribe.transcribe)(
+              processedAudio,
+              transcriptions,
+              primaryLocale,
+              {
+                ...parsed,
+                parallelism: parsed.parallelTranscribes,
+                signal: controller.signal,
+                logger,
+                onProgress: updateProgressBar
               }
+            );
+            resetProgressBar();
+            logger.info(
+              `Transcribing audiobook complete, transcriptions saved to ${transcriptions}.`
+            );
+            if (parsed.time) {
+              transcribeTiming.print();
             }
-          );
-          resetProgressBar();
-          logger.info(
-            `Transcribing audiobook complete, transcriptions saved to ${transcriptions}.`
-          );
-          if (parsed.time) {
-            transcribeTiming.print();
           }
-          const markedup = parsed.textRef === "id-fragment" ? parsed.markedup ?? (0, import_node_path.join)(os.tmpdir(), `stalign-markedup-${(0, import_node_crypto.randomUUID)()}.epub`) : parsed.epub;
+          const markedup = parsed.textRef === "id-fragment" ? parsed.markedup ?? (0, import_fs.scratchPath)(`stalign-markedup-${(0, import_node_crypto.randomUUID)()}.epub`) : parsed.epub;
           if (parsed.textRef === "id-fragment") {
             logger.info("Marking up EPUB...");
             startProgressBar();
@@ -368,11 +431,7 @@ async function main() {
               granularity: parsed.granularity,
               primaryLocale,
               logger,
-              ...!parsed.noProgress && parsed.logLevel === "silent" && {
-                onProgress: (progress) => {
-                  progressBar.update(Math.floor(progress * 100));
-                }
-              }
+              onProgress: updateProgressBar
             });
             resetProgressBar();
             logger.info(`Markup complete, marked up EPUB saved to ${markedup}.`);
@@ -384,29 +443,52 @@ async function main() {
           }
           logger.info("Aligning EPUB with audiobook...");
           startProgressBar();
-          const alignTiming = await (0, import_align.align)(
-            markedup,
-            parsed.output,
-            transcriptions,
-            processedAudio,
-            {
-              granularity: parsed.granularity,
-              textRef: parsed.textRef,
-              outFormat: parsed.outFormat,
-              primaryLocale,
-              reportsPath: parsed.reports,
-              logger,
-              ...!parsed.noProgress && parsed.logLevel === "silent" && {
-                onProgress: (progress) => {
-                  progressBar.update(Math.floor(progress * 100));
-                }
+          if ("ctc" in parsed) {
+            const timing = await (0, import_align2.ctcAlign)(
+              markedup,
+              parsed.output,
+              emissions,
+              processedAudio,
+              {
+                granularity: parsed.granularity,
+                textRef: parsed.textRef,
+                outFormat: parsed.outFormat,
+                primaryLocale: parsed.language,
+                reportsPath: parsed.reports,
+                logger,
+                onProgress: updateProgressBar
               }
+            );
+            resetProgressBar();
+            logger.info(
+              `Alignment complete, aligned EPUB saved to ${parsed.output}`
+            );
+            if (parsed.time) {
+              timing.print("Align EPUB and audiobook");
+            }
+          } else {
+            const alignTiming = await (0, import_align.align)(
+              markedup,
+              parsed.output,
+              transcriptions,
+              processedAudio,
+              {
+                granularity: parsed.granularity,
+                textRef: parsed.textRef,
+                outFormat: parsed.outFormat,
+                primaryLocale,
+                reportsPath: parsed.reports,
+                logger,
+                onProgress: updateProgressBar
+              }
+            );
+            resetProgressBar();
+            logger.info(
+              `Alignment complete, aligned EPUB saved to ${parsed.output}`
+            );
+            if (parsed.time) {
+              alignTiming.print();
             }
-          );
-          resetProgressBar();
-          logger.info(`Alignment complete, aligned EPUB saved to ${parsed.output}`);
-          if (parsed.time) {
-            alignTiming.print();
           }
           break;
         } catch (_) {
@@ -416,7 +498,15 @@ async function main() {
         }
       }
       case "snapshot": {
-        await (0, import_snapshot.snapshotAlignment)(parsed.epub, parsed.transcriptions, parsed.output);
+        if ("ctc" in parsed) {
+          await (0, import_snapshot.snapshotCtcAlignment)(parsed.epub, parsed.emissions, parsed.output);
+        } else {
+          await (0, import_snapshot2.snapshotAlignment)(
+            parsed.epub,
+            parsed.transcriptions,
+            parsed.output
+          );
+        }
       }
     }
   } catch (_2) {