@steipete/summarize 0.7.1 → 0.8.1

package/dist/esm/run/runner.js

@@ -1,8 +1,12 @@
 import { execFile } from 'node:child_process';
+import { readFile } from 'node:fs/promises';
 import { CommanderError } from 'commander';
+import { clearCacheFiles, DEFAULT_CACHE_MAX_MB, resolveCachePath, } from '../cache.js';
+import { loadSummarizeConfig } from '../config.js';
 import { parseDurationMs, parseExtractFormat, parseFirecrawlMode, parseLengthArg, parseMarkdownMode, parseMaxOutputTokensArg, parseMetricsMode, parsePreprocessMode, parseRetriesArg, parseStreamMode, parseYoutubeMode, } from '../flags.js';
 import { formatVersionLine } from '../version.js';
-import { handleHelpRequest, handleRefreshFreeRequest } from './cli-preflight.js';
+import { createCacheStateFromConfig } from './cache-state.js';
+import { handleDaemonCliRequest, handleHelpRequest, handleRefreshFreeRequest, } from './cli-preflight.js';
 import { parseCliProviderArg } from './env.js';
 import { handleFileInput, handleUrlAsset } from './flows/asset/input.js';
 import { summarizeAsset as summarizeAssetFlow } from './flows/asset/summary.js';
@@ -36,6 +40,15 @@ export async function runCli(argv, { env, fetch, execFile: execFileOverride, std
     })) {
         return;
     }
+    if (await handleDaemonCliRequest({
+        normalizedArgv,
+        envForRun,
+        fetchImpl: fetch,
+        stdout,
+        stderr,
+    })) {
+        return;
+    }
     const execFileImpl = execFileOverride ?? execFile;
     const program = buildProgram();
     program.configureOutput({
@@ -61,6 +74,82 @@ export async function runCli(argv, { env, fetch, execFile: execFileOverride, std
         stdout.write(`${formatVersionLine()}\n`);
         return;
     }
+    const promptArg = typeof program.opts().prompt === 'string' ? program.opts().prompt : null;
+    const promptFileArg = typeof program.opts().promptFile === 'string' ? program.opts().promptFile : null;
+    if (promptArg && promptFileArg) {
+        throw new Error('Use either --prompt or --prompt-file (not both).');
+    }
+    let promptOverride = null;
+    if (promptFileArg) {
+        let text;
+        try {
+            text = await readFile(promptFileArg, 'utf8');
+        }
+        catch (error) {
+            const message = error instanceof Error ? error.message : String(error);
+            throw new Error(`Failed to read --prompt-file ${promptFileArg}: ${message}`);
+        }
+        const trimmed = text.trim();
+        if (!trimmed) {
+            throw new Error(`Prompt file ${promptFileArg} is empty.`);
+        }
+        promptOverride = trimmed;
+    }
+    else if (promptArg) {
+        const trimmed = promptArg.trim();
+        if (!trimmed) {
+            throw new Error('Prompt must not be empty.');
+        }
+        promptOverride = trimmed;
+    }
+    const clearCacheFlag = normalizedArgv.includes('--clear-cache');
+    if (clearCacheFlag) {
+        const extraArgs = normalizedArgv.filter((arg) => arg !== '--clear-cache');
+        if (extraArgs.length > 0) {
+            throw new Error('--clear-cache must be used alone.');
+        }
+        const { config } = loadSummarizeConfig({ env: envForRun });
+        const cachePath = resolveCachePath({
+            env: envForRun,
+            cachePath: config?.cache?.path ?? null,
+        });
+        if (!cachePath) {
+            throw new Error('Unable to resolve cache path (missing HOME).');
+        }
+        clearCacheFiles(cachePath);
+        stdout.write('Cache cleared.\n');
+        return;
+    }
+    const cacheStatsFlag = normalizedArgv.includes('--cache-stats');
+    if (cacheStatsFlag) {
+        const extraArgs = normalizedArgv.filter((arg) => arg !== '--cache-stats');
+        if (extraArgs.length > 0) {
+            throw new Error('--cache-stats must be used alone.');
+        }
+        const { config } = loadSummarizeConfig({ env: envForRun });
+        const cachePath = resolveCachePath({
+            env: envForRun,
+            cachePath: config?.cache?.path ?? null,
+        });
+        if (!cachePath) {
+            throw new Error('Unable to resolve cache path (missing HOME).');
+        }
+        const cacheMaxMb = typeof config?.cache?.maxMb === 'number' ? config.cache.maxMb : DEFAULT_CACHE_MAX_MB;
+        const cacheMaxBytes = Math.max(0, cacheMaxMb) * 1024 * 1024;
+        const { readCacheStats } = await import('../cache.js');
+        const { formatBytes } = await import('../tty/format.js');
+        const stats = await readCacheStats(cachePath);
+        stdout.write(`Cache path: ${cachePath}\n`);
+        if (!stats) {
+            stdout.write('Cache is empty.\n');
+            return;
+        }
+        const sizeLabel = formatBytes(stats.sizeBytes);
+        const maxLabel = cacheMaxBytes > 0 ? formatBytes(cacheMaxBytes) : 'disabled';
+        stdout.write(`Size: ${sizeLabel} (max ${maxLabel})\n`);
+        stdout.write(`Entries: total=${stats.totalEntries} extract=${stats.counts.extract} summary=${stats.counts.summary} transcript=${stats.counts.transcript}\n`);
+        return;
+    }
     const cliFlagPresent = normalizedArgv.some((arg) => arg === '--cli' || arg.startsWith('--cli='));
     let cliProviderArgRaw = typeof program.opts().cli === 'string' ? program.opts().cli : null;
     const inputResolution = resolveRunInput({
@@ -84,6 +173,7 @@ export async function runCli(argv, { env, fetch, execFile: execFileOverride, std
         arg === '--lang' ||
         arg.startsWith('--lang='));
     const retries = parseRetriesArg(program.opts().retries);
+    const noCacheFlag = program.opts().cache === false;
     const extractMode = Boolean(program.opts().extract) || Boolean(program.opts().extractOnly);
     const json = Boolean(program.opts().json);
     const streamMode = parseStreamMode(program.opts().stream);
@@ -138,238 +228,283 @@ export async function runCli(argv, { env, fetch, execFile: execFileOverride, std
         cliFlagPresent,
         cliProviderArg,
     });
-    const { apiKey, openrouterApiKey, openrouterConfigured, openaiTranscriptionKey, xaiApiKey, googleApiKey, anthropicApiKey, zaiApiKey, zaiBaseUrl, firecrawlApiKey, firecrawlConfigured, googleConfigured, anthropicConfigured, apifyToken, ytDlpPath, falApiKey, cliAvailability, envForAuto, } = resolveEnvState({ env, envForRun, configForCli });
-    if (markdownModeExplicitlySet && format !== 'markdown') {
-        throw new Error('--markdown-mode is only supported with --format md');
-    }
-    if (markdownModeExplicitlySet && inputTarget.kind !== 'url') {
-        throw new Error('--markdown-mode is only supported for website URLs');
-    }
-    const metrics = createRunMetrics({
-        env,
-        fetchImpl: fetch,
-        maxOutputTokensArg,
-    });
-    const { llmCalls, trackedFetch, buildReport, estimateCostUsd, getLiteLlmCatalog, resolveMaxOutputTokensForCall, resolveMaxInputTokensForCall, setTranscriptionCost, } = metrics;
-    const { requestedModel, requestedModelInput, requestedModelLabel, isNamedModelSelection, wantsFreeNamedModel, configForModelSelection, isFallbackModel, } = resolveModelSelection({
-        config,
-        configForCli,
-        configPath,
-        envForRun,
-        explicitModelArg,
-    });
-    const verboseColor = supportsColor(stderr, envForRun);
-    const { streamingEnabled } = resolveStreamSettings({
-        streamMode,
-        stdout,
-        json,
-        extractMode,
-    });
-    if (extractMode && inputTarget.kind !== 'url') {
-        throw new Error('--extract is only supported for website/YouTube URLs');
+    if (!promptOverride && typeof config?.prompt === 'string' && config.prompt.trim().length > 0) {
+        promptOverride = config.prompt.trim();
     }
-    // Progress UI (spinner + OSC progress) is shown on stderr. Before writing to stdout (including
-    // streaming output), we stop + clear progress via the progress gate to keep scrollback clean.
-    const progressEnabled = isRichTty(stderr) && !verbose && !json;
-    const progressGate = createProgressGate();
-    const { clearProgressForStdout, setClearProgressBeforeStdout, clearProgressIfCurrent } = progressGate;
-    const fixedModelSpec = requestedModel.kind === 'fixed' ? requestedModel : null;
-    const desiredOutputTokens = resolveDesiredOutputTokens({ lengthArg, maxOutputTokensArg });
-    const summaryEngine = createSummaryEngine({
-        env,
+    const lengthInstruction = promptOverride && lengthExplicitlySet && lengthArg.kind === 'chars'
+        ? `Output is ${lengthArg.maxCharacters.toLocaleString()} characters.`
+        : null;
+    const languageInstruction = promptOverride && languageExplicitlySet && outputLanguage.kind === 'fixed'
+        ? `Output should be ${outputLanguage.label}.`
+        : null;
+    const transcriptNamespace = `yt:${youtubeMode}`;
+    const cacheState = await createCacheStateFromConfig({
         envForRun,
-        stdout,
-        stderr,
-        execFileImpl,
-        timeoutMs,
-        retries,
-        streamingEnabled,
-        plain,
-        verbose,
-        verboseColor,
-        openaiUseChatCompletions,
-        cliConfigForRun: cliConfigForRun ?? null,
-        cliAvailability,
-        trackedFetch,
-        resolveMaxOutputTokensForCall,
-        resolveMaxInputTokensForCall,
-        llmCalls,
-        clearProgressForStdout,
-        apiKeys: {
-            xaiApiKey,
-            openaiApiKey: apiKey,
-            googleApiKey,
-            anthropicApiKey,
-            openrouterApiKey,
-        },
-        keyFlags: {
-            googleConfigured,
-            anthropicConfigured,
-            openrouterConfigured,
-        },
-        zai: {
-            apiKey: zaiApiKey,
-            baseUrl: zaiBaseUrl,
-        },
+        config,
+        noCacheFlag,
+        transcriptNamespace,
     });
-    const writeViaFooter = (parts) => {
-        if (json)
-            return;
-        if (extractMode)
+    try {
+        const { apiKey, openrouterApiKey, openrouterConfigured, openaiTranscriptionKey, xaiApiKey, googleApiKey, anthropicApiKey, zaiApiKey, zaiBaseUrl, providerBaseUrls, firecrawlApiKey, firecrawlConfigured, googleConfigured, anthropicConfigured, apifyToken, ytDlpPath, falApiKey, cliAvailability, envForAuto, } = resolveEnvState({ env, envForRun, configForCli });
+        if (markdownModeExplicitlySet && format !== 'markdown') {
+            throw new Error('--markdown-mode is only supported with --format md');
+        }
+        if (markdownModeExplicitlySet && inputTarget.kind !== 'url') {
+            throw new Error('--markdown-mode is only supported for URL inputs');
+        }
+        const metrics = createRunMetrics({
+            env,
+            fetchImpl: fetch,
+            maxOutputTokensArg,
+        });
+        const { llmCalls, trackedFetch, buildReport, estimateCostUsd, getLiteLlmCatalog, resolveMaxOutputTokensForCall, resolveMaxInputTokensForCall, setTranscriptionCost, } = metrics;
+        const { requestedModel, requestedModelInput, requestedModelLabel, isNamedModelSelection, wantsFreeNamedModel, configForModelSelection, isFallbackModel, } = resolveModelSelection({
+            config,
+            configForCli,
+            configPath,
+            envForRun,
+            explicitModelArg,
+        });
+        const verboseColor = supportsColor(stderr, envForRun);
+        const { streamingEnabled } = resolveStreamSettings({
+            streamMode,
+            stdout,
+            json,
+            extractMode,
+        });
+        if (extractMode && inputTarget.kind !== 'url') {
+            throw new Error('--extract is only supported for website/YouTube URLs');
+        }
+        // Progress UI (spinner + OSC progress) is shown on stderr. Before writing to stdout (including
+        // streaming output), we stop + clear progress via the progress gate to keep scrollback clean.
+        const progressEnabled = isRichTty(stderr) && !verbose && !json;
+        const progressGate = createProgressGate();
+        const { clearProgressForStdout, setClearProgressBeforeStdout, clearProgressIfCurrent } = progressGate;
+        const fixedModelSpec = requestedModel.kind === 'fixed' ? requestedModel : null;
+        const desiredOutputTokens = resolveDesiredOutputTokens({ lengthArg, maxOutputTokensArg });
+        const summaryEngine = createSummaryEngine({
+            env,
+            envForRun,
+            stdout,
+            stderr,
+            execFileImpl,
+            timeoutMs,
+            retries,
+            streamingEnabled,
+            plain,
+            verbose,
+            verboseColor,
+            openaiUseChatCompletions,
+            cliConfigForRun: cliConfigForRun ?? null,
+            cliAvailability,
+            trackedFetch,
+            resolveMaxOutputTokensForCall,
+            resolveMaxInputTokensForCall,
+            llmCalls,
+            clearProgressForStdout,
+            apiKeys: {
+                xaiApiKey,
+                openaiApiKey: apiKey,
+                googleApiKey,
+                anthropicApiKey,
+                openrouterApiKey,
+            },
+            keyFlags: {
+                googleConfigured,
+                anthropicConfigured,
+                openrouterConfigured,
+            },
+            zai: {
+                apiKey: zaiApiKey,
+                baseUrl: zaiBaseUrl,
+            },
+            providerBaseUrls,
+        });
+        const writeViaFooter = (parts) => {
+            if (json)
+                return;
+            if (extractMode)
+                return;
+            const filtered = parts.map((p) => p.trim()).filter(Boolean);
+            if (filtered.length === 0)
+                return;
+            clearProgressForStdout();
+            stderr.write(`${ansi('2', `via ${filtered.join(', ')}`, verboseColor)}\n`);
+        };
+        const assetSummaryContext = {
+            env,
+            envForRun,
+            stdout,
+            stderr,
+            execFileImpl,
+            timeoutMs,
+            preprocessMode,
+            format,
+            lengthArg,
+            outputLanguage,
+            videoMode,
+            fixedModelSpec,
+            promptOverride,
+            lengthInstruction,
+            languageInstruction,
+            isFallbackModel,
+            desiredOutputTokens,
+            envForAuto,
+            configForModelSelection,
+            cliAvailability,
+            requestedModel,
+            requestedModelInput,
+            requestedModelLabel,
+            wantsFreeNamedModel,
+            isNamedModelSelection,
+            maxOutputTokensArg,
+            json,
+            metricsEnabled,
+            metricsDetailed,
+            shouldComputeReport,
+            runStartedAtMs,
+            verbose,
+            verboseColor,
+            streamingEnabled,
+            plain,
+            summaryEngine,
+            trackedFetch,
+            writeViaFooter,
+            clearProgressForStdout,
+            getLiteLlmCatalog,
+            buildReport,
+            estimateCostUsd,
+            llmCalls,
+            cache: cacheState,
+            apiStatus: {
+                xaiApiKey,
+                apiKey,
+                openrouterApiKey,
+                apifyToken,
+                firecrawlConfigured,
+                googleConfigured,
+                anthropicConfigured,
+                providerBaseUrls,
+                zaiApiKey,
+                zaiBaseUrl,
+            },
+        };
+        const summarizeAsset = (args) => summarizeAssetFlow(assetSummaryContext, args);
+        const assetInputContext = {
+            env,
+            stderr,
+            progressEnabled,
+            timeoutMs,
+            trackedFetch,
+            summarizeAsset,
+            setClearProgressBeforeStdout,
+            clearProgressIfCurrent,
+        };
+        if (await handleFileInput(assetInputContext, inputTarget)) {
             return;
-        const filtered = parts.map((p) => p.trim()).filter(Boolean);
-        if (filtered.length === 0)
+        }
+        if (url && (await handleUrlAsset(assetInputContext, url, isYoutubeUrl))) {
             return;
-        clearProgressForStdout();
-        stderr.write(`${ansi('2', `via ${filtered.join(', ')}`, verboseColor)}\n`);
-    };
-    const assetSummaryContext = {
-        env,
-        envForRun,
-        stdout,
-        stderr,
-        execFileImpl,
-        timeoutMs,
-        preprocessMode,
-        format,
-        lengthArg,
-        outputLanguage,
-        videoMode,
-        fixedModelSpec,
-        isFallbackModel,
-        desiredOutputTokens,
-        envForAuto,
-        configForModelSelection,
-        cliAvailability,
-        requestedModel,
-        requestedModelInput,
-        requestedModelLabel,
-        wantsFreeNamedModel,
-        isNamedModelSelection,
-        maxOutputTokensArg,
-        json,
-        metricsEnabled,
-        metricsDetailed,
-        shouldComputeReport,
-        runStartedAtMs,
-        verbose,
-        verboseColor,
-        streamingEnabled,
-        plain,
-        summaryEngine,
-        trackedFetch,
-        writeViaFooter,
-        clearProgressForStdout,
-        getLiteLlmCatalog,
-        buildReport,
-        estimateCostUsd,
-        llmCalls,
-        apiStatus: {
-            xaiApiKey,
-            apiKey,
-            openrouterApiKey,
-            apifyToken,
-            firecrawlConfigured,
-            googleConfigured,
-            anthropicConfigured,
-            zaiApiKey,
-            zaiBaseUrl,
-        },
-    };
-    const summarizeAsset = (args) => summarizeAssetFlow(assetSummaryContext, args);
-    const assetInputContext = {
-        env,
-        stderr,
-        progressEnabled,
-        timeoutMs,
-        trackedFetch,
-        summarizeAsset,
-        setClearProgressBeforeStdout,
-        clearProgressIfCurrent,
-    };
-    if (await handleFileInput(assetInputContext, inputTarget)) {
-        return;
-    }
-    if (url && (await handleUrlAsset(assetInputContext, url, isYoutubeUrl))) {
-        return;
+        }
+        if (!url) {
+            throw new Error('Only HTTP and HTTPS URLs can be summarized');
+        }
+        const urlFlowContext = {
+            io: {
+                env,
+                envForRun,
+                stdout,
+                stderr,
+                execFileImpl,
+                fetch: trackedFetch,
+            },
+            flags: {
+                timeoutMs,
+                maxExtractCharacters: null,
+                retries,
+                format,
+                markdownMode,
+                preprocessMode,
+                youtubeMode,
+                firecrawlMode: requestedFirecrawlMode,
+                videoMode,
+                outputLanguage,
+                lengthArg,
+                promptOverride,
+                lengthInstruction,
+                languageInstruction,
+                maxOutputTokensArg,
+                json,
+                extractMode,
+                metricsEnabled,
+                metricsDetailed,
+                shouldComputeReport,
+                runStartedAtMs,
+                verbose,
+                verboseColor,
+                progressEnabled,
+                streamingEnabled,
+                plain,
+                configPath,
+                configModelLabel,
+            },
+            model: {
+                requestedModel,
+                requestedModelInput,
+                requestedModelLabel,
+                fixedModelSpec,
+                isFallbackModel,
+                isNamedModelSelection,
+                wantsFreeNamedModel,
+                desiredOutputTokens,
+                configForModelSelection,
+                envForAuto,
+                cliAvailability,
+                openaiUseChatCompletions,
+                openaiWhisperUsdPerMinute,
+                apiStatus: {
+                    xaiApiKey,
+                    apiKey,
+                    openrouterApiKey,
+                    openrouterConfigured,
+                    googleApiKey,
+                    googleConfigured,
+                    anthropicApiKey,
+                    anthropicConfigured,
+                    providerBaseUrls,
+                    zaiApiKey,
+                    zaiBaseUrl,
+                    firecrawlConfigured,
+                    firecrawlApiKey,
+                    apifyToken,
+                    ytDlpPath,
+                    falApiKey,
+                    openaiTranscriptionKey,
+                },
+                summaryEngine,
+                getLiteLlmCatalog,
+                llmCalls,
+            },
+            cache: cacheState,
+            hooks: {
+                onModelChosen: null,
+                onExtracted: null,
+                onLinkPreviewProgress: null,
+                onSummaryCached: null,
+                setTranscriptionCost,
+                summarizeAsset,
+                writeViaFooter,
+                clearProgressForStdout,
+                setClearProgressBeforeStdout,
+                clearProgressIfCurrent,
+                buildReport,
+                estimateCostUsd,
+            },
+        };
+        await runUrlFlow({ ctx: urlFlowContext, url, isYoutubeUrl });
     }
-    if (!url) {
-        throw new Error('Only HTTP and HTTPS URLs can be summarized');
+    finally {
+        cacheState.store?.close();
     }
-    const urlFlowContext = {
-        env,
-        envForRun,
-        stdout,
-        stderr,
-        execFileImpl,
-        timeoutMs,
-        retries,
-        format,
-        markdownMode,
-        preprocessMode,
-        youtubeMode,
-        firecrawlMode: requestedFirecrawlMode,
-        videoMode,
-        outputLanguage,
-        lengthArg,
-        maxOutputTokensArg,
-        requestedModel,
-        requestedModelInput,
-        requestedModelLabel,
-        fixedModelSpec,
-        isFallbackModel,
-        isNamedModelSelection,
-        wantsFreeNamedModel,
-        desiredOutputTokens,
-        configForModelSelection,
-        envForAuto,
-        cliAvailability,
-        json,
-        extractMode,
-        metricsEnabled,
-        metricsDetailed,
-        shouldComputeReport,
-        runStartedAtMs,
-        verbose,
-        verboseColor,
-        progressEnabled,
-        streamingEnabled,
-        plain,
-        openaiUseChatCompletions,
-        configPath,
-        configModelLabel,
-        openaiWhisperUsdPerMinute,
-        setTranscriptionCost,
-        apiStatus: {
-            xaiApiKey,
-            apiKey,
-            openrouterApiKey,
-            openrouterConfigured,
-            googleApiKey,
-            googleConfigured,
-            anthropicApiKey,
-            anthropicConfigured,
-            zaiApiKey,
-            zaiBaseUrl,
-            firecrawlConfigured,
-            firecrawlApiKey,
-            apifyToken,
-            ytDlpPath,
-            falApiKey,
-            openaiTranscriptionKey,
-        },
-        trackedFetch,
-        summaryEngine,
-        summarizeAsset,
-        writeViaFooter,
-        clearProgressForStdout,
-        setClearProgressBeforeStdout,
-        clearProgressIfCurrent,
-        getLiteLlmCatalog,
-        buildReport,
-        estimateCostUsd,
-        llmCalls,
-    };
-    await runUrlFlow({ ctx: urlFlowContext, url, isYoutubeUrl });
 }
 //# sourceMappingURL=runner.js.map