Working on a more efficient manifest reader.

frontend/src/js/component/wb-collection-content.js

@@ -2,6 +2,7 @@ import { h, Component } from 'preact';
 import WBTable from 'wb-table';
 import WBBreadcrumbs from 'wb-breadcrumbs';
 import { WBManifestReader } from 'wb-collection-manifest';
+// import WBManifestReader from 'wb-manifest-reader';
 import WBPagination from 'wb-pagination';
 import makeArvadosRequest from 'make-arvados-request';
 import wbDownloadFile from 'wb-download-file';

frontend/src/js/misc/wb-manifest-reader.js

@@ -0,0 +1,127 @@
+function mkdir(parent, name) {
+  if (name in parent && (parent[name] instanceof Array))
+    throw Error('File with the same name already exists');
+  if (name in parent)
+    return parent[name];
+  const dir = {};
+  parent[name] = dir;
+  return dir;
+}
+
+function mkpath(parent, path) {
+  if (typeof(path) === 'string')
+    path = path.split('/');
+  let dir = parent;
+  for (let i = 1; i < path.length; i++) {
+    dir = mkdir(dir, path[i]);
+  }
+  return dir;
+}
+
+function appendFile(dir, name, sidx, seg) {
+  if (name in dir && (!(dir[name] instanceof Array)))
+    throw Error('Directory with the same name already exists');
+  if (!(name in dir))
+    dir[name] = [];
+  const f = dir[name];
+  f.push([ sidx, seg[0], seg[1] ]);
+  return f;
+}
+
+function process(streams) {
+  const rootDir = {};
+
+  streams.map((s, sidx) => {
+    const [ streamName, locators, segments ] = s;
+    const streamDir = mkpath(rootDir, streamName);
+    segments.map((seg, segidx) => {
+      let name = seg[2].split('/');
+      const dir = (name.length === 1 ? streamDir :
+        mkpath(streamDir, ['.'].concat(name.slice(0, name.length - 1))));
+      name = name[name.length - 1];
+      appendFile(dir, name, sidx, seg);
+    });
+  });
+
+  return rootDir;
+}
+
+function parse(manifestText) {
+  const M_STREAM_NAME = 0;
+  const M_LOCATORS = 1;
+  const M_FILE_SEGMENTS = 2;
+
+  let mode = M_STREAM_NAME;
+
+  const streams = [];
+  let locators = [];
+  let segments = [];
+
+  let streamName;
+  let accum = '';
+  let tokenStart = 0;
+
+  for (let i = 0; i < manifestText.length; i++) {
+    const c = manifestText[i];
+
+    if (mode === M_STREAM_NAME) {
+      if (c === ' ') {
+        mode = M_LOCATORS;
+        streamName = accum;
+        accum = '';
+        tokenStart = i + 1;
+      } else {
+        accum += c;
+      }
+
+    } else if (mode === M_LOCATORS) {
+      if (c === ':') {
+        mode = M_FILE_SEGMENTS;
+        accum = '';
+        i = tokenStart - 1;
+        let pos = 0;
+        locators = locators.map(loc => {
+          const r = loc.concat([ pos, pos + loc[1] ]);
+          pos += loc[1];
+          return r;
+        });
+      } else if (c === ' ') {
+        const sz = Number(accum.split('+')[1]);
+        locators.push([accum, sz]);
+        accum = '';
+        tokenStart = i + 1;
+      } else {
+        accum += c;
+      }
+
+    } else if (mode === M_FILE_SEGMENTS) {
+      if (c === ' ' || c === '\n') {
+        let seg = accum.split(':');
+        seg = [Number(seg[0]), Number(seg[1]), seg[2]];
+        segments.push(seg);
+        accum = '';
+        tokenStart = i + 1;
+        if (c === '\n') {
+          streams.push([streamName, locators, segments]);
+          locators = [];
+          segments = [];
+          mode = M_STREAM_NAME;
+        }
+      } else {
+        accum += c;
+      }
+
+    }
+  }
+
+  return streams;
+}
+
+class WBManifestReader {
+  constructor(manifestText) {
+    this.streams = parse(manifestText);
+    this.rootDir = process(this.streams);
+  }
+}
+
+export default WBManifestReader;