Created
August 1, 2012 06:25
-
-
Save trentm/3224233 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Index: /Users/trentm/tm/json/lib/jsontool.js | |
index 734729c..d2764f9 100755 | |
--- a/lib/jsontool.js | |
+++ b/lib/jsontool.js | |
@@ -399,36 +399,140 @@ function parseArgv(argv) { | |
* Get input from either given file paths or stdin. | |
* | |
* @param opts {Object} Parsed options. | |
- * @param callback {Function} `function (err, callback)` where err is an | |
- * error string if there was a problem. | |
+ * @param callback {Function} `function (err, chunk)` where err is an | |
+ * error string if there was a problem. This is called once for each | |
+ * "chunk". XXX START HERE: splain chunks | |
*/ | |
-function getInput(opts, callback) { | |
- if (opts.inputFiles.length === 0) { | |
- // Read from stdin. | |
- var chunks = []; | |
- | |
- var stdin = process.openStdin(); | |
- stdin.setEncoding('utf8'); | |
- stdin.on('data', function (chunk) { | |
- chunks.push(chunk); | |
- }); | |
+function jsonChunksFromInput(opts, callback) { | |
+ // If returned from `stripHeaders()` and `finishedHeaders` is still false, | |
+ // then we've process a chunk with an incomplete set of headers: | |
+ // `stripHeaders()` should be called again with the next chunk. | |
+ var finishedHeaders = false; | |
+ function stripHeaders(s) { | |
+ // Take off a leading HTTP header if any and pass it through. | |
+ while (true) { | |
+ if (s.slice(0,5) === "HTTP/") { | |
+ var index = s.indexOf('\r\n\r\n'); | |
+ var sepLen = 4; | |
+ if (index == -1) { | |
+ index = s.indexOf('\n\n'); | |
+ sepLen = 2; | |
+ } | |
+ if (index != -1) { | |
+ if (! opts.dropHeaders) { | |
+ emit(s.slice(0, index+sepLen)); | |
+ } | |
+ var is100Continue = (s.slice(0, 21) === "HTTP/1.1 100 Continue"); | |
+ s = s.slice(index+sepLen); | |
+ if (is100Continue) { | |
+ continue; | |
+ } | |
+ finishedHeaders = true; | |
+ } | |
+ } else { | |
+ finishedHeaders = true; | |
+ } | |
+ break; | |
+ } | |
+ //console.warn("XXX stripHeaders done, finishedHeaders=%s", finishedHeaders) | |
+ return s; | |
+ } | |
- stdin.on('end', function () { | |
- callback(null, chunks.join('')); | |
- }); | |
- } else { | |
+ if (opts.inputFiles.length > 0) { | |
// Read input files. | |
+ // TODO: Improve streaming here: read files async in chunks and stream | |
+ // as above if `-ga`. | |
var i = 0; | |
var chunks = []; | |
try { | |
- for (; i < opts.inputFiles.length; i++) { | |
- chunks.push(fs.readFileSync(opts.inputFiles[i], 'utf8')); | |
+ var first = fs.readFileSync(opts.inputFiles[i], 'utf8'); | |
+ first = stripHeaders(first); | |
+ callback(null, first); | |
+ for (i++; i < opts.inputFiles.length; i++) { | |
+ callback(null, fs.readFileSync(opts.inputFiles[i], 'utf8')); | |
} | |
} catch (e) { | |
return callback( | |
format('could not read "%s": %s', opts.inputFiles[i], e)); | |
} | |
- callback(null, chunks.join('')); | |
+ } else if (opts.group && opts.array && opts.outputMode !== OM_JSON) { | |
+ // Streaming from stdin. | |
+ //console.warn("XXX streaming"); | |
+ var streaming = true; | |
+ var leftover = ''; | |
+ var chunks = []; | |
+ var splitter = /(})(\s*\n\s*)?({\s*")/; | |
+ function callbackJsonChunks(chunk) { | |
+ if (chunk[0] !== '{') { // Only support streaming consecutive *objects*. | |
+ streaming = false; | |
+ chunks.push(chunk); | |
+ return; | |
+ } | |
+ /* Example: | |
+ * > '{"a":"b"}\n{"a":"b"}\n{"a":"b"}'.split(/(})(\s*\n\s*)?({\s*")/) | |
+ * [ '{"a":"b"', | |
+ * '}', | |
+ * '\n', | |
+ * '{"', | |
+ * 'a":"b"', | |
+ * '}', | |
+ * '\n', | |
+ * '{"', | |
+ * 'a":"b"}' ] | |
+ */ | |
+ var bits = chunk.split(splitter); | |
+ //console.warn("XXX bits: ", bits) | |
+ if (bits.length === 1) { | |
+ leftover = chunk; | |
+ } else { | |
+ var n = bits.length - 2; | |
+ callback(null, bits[0] + bits[1]); | |
+ for (var i = 3; i < n; i += 4) { | |
+ callback(null, bits[i] + bits[i+1] + bits[i+2]); | |
+ } | |
+ leftover = bits[n] + bits[n+1]; | |
+ } | |
+ } | |
+ | |
+ var stdin = process.openStdin(); | |
+ stdin.setEncoding('utf8'); | |
+ stdin.on('data', function (chunk) { | |
+ //console.warn("XXX process chunk: %s", JSON.stringify(chunk)) | |
+ if (!streaming) { | |
+ chunks.push(chunk); | |
+ return; | |
+ } | |
+ var s = leftover + chunk; | |
+ if (!finishedHeaders) { | |
+ s = stripHeaders(s); | |
+ } | |
+ if (!finishedHeaders) { | |
+ leftover = s; | |
+ } else { | |
+ callbackJsonChunks(s); | |
+ } | |
+ }); | |
+ stdin.on('end', function () { | |
+ if (!streaming) { | |
+ callback(null, chunks.join('')); | |
+ } else if (leftover) { | |
+ callbackJsonChunks(leftover); | |
+ callback(null, leftover); | |
+ } | |
+ }); | |
+ } else { | |
+ // Read stdin in one big chunk. | |
+ var stdin = process.openStdin(); | |
+ stdin.setEncoding('utf8'); | |
+ var chunks = []; | |
+ stdin.on('data', function (chunk) { | |
+ chunks.push(chunk); | |
+ }); | |
+ stdin.on('end', function () { | |
+ var chunk = chunks.join(''); | |
+ chunk = stripHeaders(chunk); | |
+ callback(null, chunks.join('')); | |
+ }); | |
} | |
} | |
@@ -590,6 +694,7 @@ function parseInput(buffer, group, merge) { | |
// This condition should be fine for typical use cases and ensures | |
// no false matches inside JS strings. | |
var newBuffer = buffer; | |
+ //XXX START HERE | |
[/(})\s*\n\s*({)/g, /(})({")/g].forEach(function (pat) { | |
newBuffer = newBuffer.replace(pat, "$1,\n$2"); | |
}); | |
@@ -820,55 +925,33 @@ function main(argv) { | |
} | |
var lookupStrs = opts.args; | |
- getInput(opts, function (err, buffer) { | |
+ jsonChunksFromInput(opts, function (err, chunk) { | |
+ //console.warn("XXX chunk: '%s'", chunk) | |
if (err) { | |
warn("json: error: %s", err) | |
return drainStdoutAndExit(1); | |
} | |
- // Take off a leading HTTP header if any and pass it through. | |
- while (true) { | |
- if (buffer.slice(0,5) === "HTTP/") { | |
- var index = buffer.indexOf('\r\n\r\n'); | |
- var sepLen = 4; | |
- if (index == -1) { | |
- index = buffer.indexOf('\n\n'); | |
- sepLen = 2; | |
- } | |
- if (index != -1) { | |
- if (! opts.dropHeaders) { | |
- emit(buffer.slice(0, index+sepLen)); | |
- } | |
- var is100Continue = (buffer.slice(0, 21) === "HTTP/1.1 100 Continue"); | |
- buffer = buffer.slice(index+sepLen); | |
- if (is100Continue) { | |
- continue; | |
- } | |
- } | |
- } | |
- break; | |
- } | |
- | |
// Expect the remainder to be JSON. | |
- if (! buffer.length) { | |
+ if (! chunk.length) { | |
return; | |
} | |
// parseInput() -> {datum: <input object>, error: <error object>} | |
- var input = parseInput(buffer, opts.group, opts.merge); | |
+ var input = parseInput(chunk, opts.group, opts.merge); | |
if (input.error) { | |
// Doesn't look like JSON. Just print it out and move on. | |
if (! opts.quiet) { | |
// Use JSON-js' "json_parse" parser to get more detail on the | |
// syntax error. | |
var details = ""; | |
- var normBuffer = buffer.replace(/\r\n|\n|\r/, '\n'); | |
+ var normBuffer = chunk.replace(/\r\n|\n|\r/, '\n'); | |
try { | |
json_parse(normBuffer); | |
details = input.error; | |
} catch(err) { | |
// err.at has the position. Get line/column from that. | |
var at = err.at - 1; // `err.at` looks to be 1-based. | |
- var lines = buffer.split('\n'); | |
+ var lines = chunk.split('\n'); | |
var line, col, pos = 0; | |
for (line = 0; line < lines.length; line++) { | |
pos += lines[line].length + 1; | |
@@ -887,8 +970,8 @@ function main(argv) { | |
warn("json: error: input is not JSON: %s", details); | |
} | |
if (!opts.validate) { | |
- emit(buffer); | |
- if (buffer.length && buffer[buffer.length-1] !== "\n") { | |
+ emit(chunk); | |
+ if (chunk.length && chunk[chunk.length-1] !== "\n") { | |
emit('\n'); | |
} | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment