aboutsummaryrefslogtreecommitdiffstats
path: root/libs
diff options
context:
space:
mode:
authorMatěj Cepl <mcepl@redhat.com>2013-05-14 02:10:40 +0200
committerMatěj Cepl <mcepl@redhat.com>2013-05-18 20:57:39 +0200
commita9e1cfc3c94879480862d3ef764f4fd7fed57d4a (patch)
tree550cb7fdaf2abf897668d2718eecd0d257370d8c /libs
parentf395dfeb002d0840b5e92d23158801d15d632d9e (diff)
downloadimportLDIF-a9e1cfc3c94879480862d3ef764f4fd7fed57d4a.tar.gz
Add OpenwebApps packaged app parafernalia
Diffstat (limited to 'libs')
-rw-r--r--[l---------]libs/parseLDIF.js296
1 files changed, 295 insertions, 1 deletions
diff --git a/libs/parseLDIF.js b/libs/parseLDIF.js
index 0df7340..e060f54 120000..100644
--- a/libs/parseLDIF.js
+++ b/libs/parseLDIF.js
@@ -1 +1,295 @@
-parseLDIF/parseLDIF.js \ No newline at end of file
+// Compatibility shims
+if (!String.prototype.trim) {
+ String.prototype.trim = function () {
+ return this.replace(/^\s+|\s+$/g,'');
+ };
+}
+
+if (!String.prototype.trimRight) {
+ String.prototype.trimRight = function () {
+ return this.replace(/\s+$/,'');
+ };
+}
+
+if (!Array.isArray) {
+ Array.isArray = function (vArg) {
+ return Object.prototype.toString.call(vArg) === "[object Array]";
+ };
+}
+
+const debugState = false;
+
+/**
+*
+* Base64 encode / decode
+* http://www.webtoolkit.info/
+* From http://stackoverflow.com/questions/246801\
+* /how-can-you-encode-to-base64-using-javascript
+*
+**/
+var Base64 = {
+
+ // private property
+ _keyStr : "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=",
+
+ // public method for encoding
+ encode : function (input) {
+ var output = "";
+ var chr1, chr2, chr3, enc1, enc2, enc3, enc4;
+ var i = 0;
+
+ input = Base64._utf8_encode(input);
+
+ while (i < input.length) {
+
+ chr1 = input.charCodeAt(i++);
+ chr2 = input.charCodeAt(i++);
+ chr3 = input.charCodeAt(i++);
+
+ enc1 = chr1 >> 2;
+ enc2 = ((chr1 & 3) << 4) | (chr2 >> 4);
+ enc3 = ((chr2 & 15) << 2) | (chr3 >> 6);
+ enc4 = chr3 & 63;
+
+ if (isNaN(chr2)) {
+ enc3 = enc4 = 64;
+ } else if (isNaN(chr3)) {
+ enc4 = 64;
+ }
+
+ output = output +
+ this._keyStr.charAt(enc1) + this._keyStr.charAt(enc2) +
+ this._keyStr.charAt(enc3) + this._keyStr.charAt(enc4);
+
+ }
+
+ return output;
+ },
+
+ // public method for decoding
+ decode : function (input) {
+ var output = "";
+ var chr1, chr2, chr3;
+ var enc1, enc2, enc3, enc4;
+ var i = 0;
+
+ input = input.replace(/[^A-Za-z0-9\+\/\=]/g, "");
+
+ while (i < input.length) {
+
+ enc1 = this._keyStr.indexOf(input.charAt(i++));
+ enc2 = this._keyStr.indexOf(input.charAt(i++));
+ enc3 = this._keyStr.indexOf(input.charAt(i++));
+ enc4 = this._keyStr.indexOf(input.charAt(i++));
+
+ chr1 = (enc1 << 2) | (enc2 >> 4);
+ chr2 = ((enc2 & 15) << 4) | (enc3 >> 2);
+ chr3 = ((enc3 & 3) << 6) | enc4;
+
+ output = output + String.fromCharCode(chr1);
+
+ if (enc3 != 64) {
+ output = output + String.fromCharCode(chr2);
+ }
+ if (enc4 != 64) {
+ output = output + String.fromCharCode(chr3);
+ }
+
+ }
+
+ output = Base64._utf8_decode(output);
+
+ return output;
+
+ },
+
+ // private method for UTF-8 encoding
+ _utf8_encode : function (string) {
+ string = string.replace(/\r\n/g,"\n");
+ var utftext = "";
+
+ for (var n = 0; n < string.length; n++) {
+
+ var c = string.charCodeAt(n);
+
+ if (c < 128) {
+ utftext += String.fromCharCode(c);
+ }
+ else if((c > 127) && (c < 2048)) {
+ utftext += String.fromCharCode((c >> 6) | 192);
+ utftext += String.fromCharCode((c & 63) | 128);
+ }
+ else {
+ utftext += String.fromCharCode((c >> 12) | 224);
+ utftext += String.fromCharCode(((c >> 6) & 63) | 128);
+ utftext += String.fromCharCode((c & 63) | 128);
+ }
+
+ }
+
+ return utftext;
+ },
+
+ // private method for UTF-8 decoding
+ _utf8_decode : function (utftext) {
+ var string = "";
+ var i = 0;
+ var c = c1 = c2 = 0;
+
+ while ( i < utftext.length ) {
+
+ c = utftext.charCodeAt(i);
+
+ if (c < 128) {
+ string += String.fromCharCode(c);
+ i++;
+ }
+ else if((c > 191) && (c < 224)) {
+ c2 = utftext.charCodeAt(i+1);
+ string += String.fromCharCode(((c & 31) << 6) | (c2 & 63));
+ i += 2;
+ }
+ else {
+ c2 = utftext.charCodeAt(i+1);
+ c3 = utftext.charCodeAt(i+2);
+ string += String.fromCharCode(((c & 15) << 12) | ((c2 & 63) << 6) | (c3 & 63));
+ i += 3;
+ }
+
+ }
+
+ return string;
+ }
+
+}
+
+const usefulFields = ["birthyear", "c", "cn", "description",
+ "facsimiletelephonenumber", "givenName", "homePhone", "l", "mail",
+ "mobile", "mozillaHomeCountryName", "mozillaHomeLocalityName",
+ "mozillaHomePostalCode", "mozillaHomeState", "mozillaHomeStreet",
+ "mozillaHomeUrl", "mozillaNickname", "o", "sn", "st", "street",
+ "telephoneNumber", "title", "objectclass"
+ ];
+
+function debug(str) {
+ if (debugState) {
+ print(str);
+ }
+}
+
+/**
+ * parse LDIF string into JavaScript Object
+ *
+ * @inStr String
+ * @returns Object
+ *
+ * @todo
+ * Tries to follow RFC2849, but
+ * - doesn’t even consider change requests, only complete values
+ * - doesn’t include version: field (seems to be ignored by
+ * Thunderbird as well).
+ * - ignores multi-value attributes completely
+ * - ignores < links
+ */
+function parseLDIF(inStr) {
+ var record = {},
+ key = "",
+ value = null,
+ splitLine = [],
+ colon_idx = 0,
+ out_records = [];
+
+ function handleAdding(key, value) {
+ value = value ? value.trim() : "";
+
+ // base64 encoded value
+ if (value[0] === ":") {
+ value = Base64.decode(value.slice(1).trim());
+ }
+
+ if (key && (usefulFields.indexOf(key) != -1) &&
+ value.length > 0) {
+ if (key in this) {
+ if (Array.isArray(this[key])) {
+ this[key].push(value);
+ }
+ else {
+ this[key] = new Array(this[key]);
+ this[key].push(value);
+ }
+ }
+ else {
+ this[key] = value;
+ }
+ }
+
+ key = "";
+ value = null;
+ }
+
+ record.add = handleAdding;
+
+ inStr.forEach(function (line) {
+ if (line != undefined) {
+ line = line.trim();
+
+ if (line.length == 0) {
+ // > 1, because we have always .add property
+ if (Object.keys(record).length > 1) {
+ record.add(key, value);
+ delete record.add;
+
+ if (record.objectclass &&
+ (record.objectclass === "person" ||
+ record.objectclass.indexOf("person") !== -1)) {
+ delete record.objectclass;
+ out_records.push(record);
+ }
+ }
+ record = {};
+ record.add = handleAdding;
+ key = "";
+ value = null;
+ }
+ else {
+ // comment line
+ if (line[0] === "#") {
+ return ;
+ }
+ else {
+ colon_idx = line.indexOf(":");
+ line = line.trim();
+
+ // Line continuation
+ // This is how it works in THunderbird exported LDIF
+ // files, but not what RFC describes ... the first character
+ // of continuing entry should be space, and the first line
+ // of value should never be empty. Hopefully, this solution
+ // should be at least compatible for reading of RFC LDIF
+ // files (so it doesn’t hurt)
+ if (colon_idx == -1) {
+ // multiline value
+ if (line[0] === " ") {
+ line = line.slice(1);
+ }
+ value += line;
+ }
+ else {
+ record.add(key, value);
+
+ key = line.slice(0, colon_idx);
+ value = line.slice(colon_idx + 1);
+ }
+
+ }
+ }
+ }
+ });
+
+ return out_records;
+}
+
+if (arguments && arguments.length == 1) {
+ var lines = readFile(arguments[0]).replace(/\r\n/g,"\n");
+ print(parseLDIF(lines.split("\n")).toSource());
+}