contentscript.js

// Main script
//
// Turns #include statments into hyperlinks
// Also turns python import/process.load() statements into links as well
//
// Robin Aggleton 2014


// console.log("Debugging github-linkify-cmssw");


// make root URL, i.e. up to & including CMSSW_X_Y_Z_AAA, or tag hash
// note, be careful as can have URLs in diff formats:
// github.com/cms-sw/cmssw/blob/CMSSW_7_2_X/DataFormats/JetReco/interface/GenJetCollection.h
// with CMSSW_1_2_3, CMSSW_1_2_X, CMSSW_1_2_3_SHLC_fixTk, l1t-devel-CMSSW_7_3_0_pre2, etc
// or
// github.com/cms-sw/cmssw/blob/975...a78/DataFormats/JetReco/interface/GenJetCollection.h
var url = document.URL;
var cmssw_pattern = /.*\/cmssw\/blob\/[_a-zA-Z0-9\.\-]*?\//;
var rootURL = "";
cmssw = cmssw_pattern.exec(url)[0];
rootURL = url.substr(0, url.search(cmssw_pattern)+cmssw.length);
// console.log("URL: " + rootURL);

// test line to see if is a valid "#include"
function test_cpp_include(line) {
    // var cpp_pattern = /#include/;
    if (/#include/.test(line)
        && (/\.h/.test(line))
        && (line.indexOf("/") > 5) // to eliminate STL, the > 2 is to avoid commented lines as well
        && (line.search("boost") == -1) // ignore popular libs
        )
        return true;
    else
        return false
}


// test line to see if valid python import/process.load/etc
// returns sanitized python part of line (e.g. RecoJets.config.blah)
function test_py_import(line) {
    // patterns to do python matching
    var py_pattern1 = /from\s.*\simport\s.*/;
    var py_pattern2 = /import.*as.*/;
    var py_pattern3 = /import/;
    var py_pattern4 = /process.load(.*)/;

    line = line.trim();

    var config;
    if (py_pattern1.test(line)) {
        config = line.replace(/from\s/,"");
        config = config.replace(/\simport.*/,"");
    } else if (py_pattern2.test(line)) {
        config = line.replace(/import\s/,"");
        config = config.replace(/\sas\s.*/,"");
    } else if (py_pattern3.test(line)) {
        config = line.replace(/import\s/,"");
    } else if (py_pattern4.test(line)) {
        config = line.replace("process.load(","");
        config = config.replace(")","");
    }
    // ensure it has a / in it so it isn't a 3rd party library
    if ((/\./.test(config) || /\//.test(config)))
        return config;
    else
        return "";
}


// test if URL exists
// make sure it's asynchronous otherwise hangs!
function check_URL_add_link(url, cell) {
    var http = new XMLHttpRequest();
    http.open('HEAD', url);
    http.onreadystatechange = function() {
        if (this.readyState == 4) {
            if (this.status != 404) {
                var srcText = "&nbsp&nbsp&nbsp<a href=\""+url+"\" style=\"text-decoration:underline;\">[goto src]</a>";
                cell.innerHTML = cell.innerHTML.concat(srcText);
            }
        }
    };
    http.send();
}


// loop through all rows <tr>
var rows = document.getElementsByTagName("tr");
for (var i = 0; i < rows.length; ++i) {
    for (var j = 0; j < rows[i].childNodes.length; ++j) {

        // check it's a line of the file, not anything else
        var file_line_pattern = /js-file-line/;
        var cell = rows[i].childNodes[j];

        if (file_line_pattern.test(cell.className)) {
            var line = cell.textContent;

            /////////////
            // C++ BIT //
            /////////////
            // check line has a C++ #include, if so pull the header path
            // ensure it has a ".h", and a "/", to exclude STL
            if (test_cpp_include(line)) {
                var cpp_header_pattern = /[<\"].*[>\"]/; // does "myheader.h" or <myheader.h>
                var path = cpp_header_pattern.exec(line)[0];
                path = path.replace(/[<>\"]/g, ""); //g inmportant - global so replaces all instances
                var link = rootURL.concat(path);

                // get color of original text by getting class name of <span> or <td> tag that houses it,
                // then ask style sheet for color attribute for that class
                var header_class = "";
                for (var k = 0; k < cell.childNodes.length; k++) {
                    if (cell.childNodes[k].textContent.indexOf(path) != -1) {
                        header_class = cell.childNodes[k].className;
                    }
                }
                if (!header_class) {
                    header_class = cell.className.split(" ")[0];
                }

                var color = window.getComputedStyle(document.querySelector("."+header_class)).getPropertyValue("color");

                // let's replace the text with a link
                // keep same colour as before, but underline it to make it noticeable for user
                var newText = "<a href=\""+link+"\" style=\"text-decoration:underline;color:"+color+"\">"+path+"</a>";
                cell.innerHTML = cell.innerHTML.replace(path, newText); // <font color=\"#df5000\">

                // we add in a tag to go to the src file instead
                // THIS ASSUMES THE FILE IS IN /src/ DIRECTORY
                // but what if it is in /plugins???
                var srcPath = link.replace("interface", "src");
                var cppSrc = srcPath.replace(".h", ".cpp");
                var ccSrc = srcPath.replace(".h", ".cc");
                // try .cc first, if not try cpp
                check_URL_add_link(ccSrc, cell);
                check_URL_add_link(cppSrc, cell);
            }

            ////////////////
            // PYTHON BIT //
            ////////////////
            // tests for python imports/fragments
            var pyImportPart = test_py_import(line);
            // if we have a valid python config, turn it into a path
            if (pyImportPart != "") {
                pyImportPart = pyImportPart.replace(/['"]/g, ""); // remove quotations
                var path = pyImportPart.replace(/\./g, "/"); // replace . with /
                var parts = path.split("/");
                // Now insert the missing 'python' directory
                // Need to handle possible directories inside of python dir
                // In CMSSW python is assumed to be AAA/BBB/python,
                // so that AAA/BBB/python/CCC is OK
                // but AAA/BBB/CCC/python is not OK
                // path = path.replace(parts[2],"python/"+parts[2]+".py")
                parts.splice(2, 0, "python");
                // Add in the .py ending
                parts.splice(-1, 1, parts.slice(-1)[0] + ".py");
                // Turn back into string
                path = parts.join("/");
                var link = rootURL.concat(path)

                // get color of original text by getting class name of <span> or <td> tag that houses it,
                // then ask style sheet for color attribute for that class
                // should probably put this as a function since we do the same for py and c++
                var header_class = "";
                for (var k = 0; k < cell.childNodes.length; k++) {
                    if (cell.childNodes[k].textContent.indexOf(pyImportPart) != -1) {
                        header_class = cell.childNodes[k].className;
                    }
                }
                if (!header_class){
                    header_class = cell.className.split(" ")[0];
                }
                var color = window.getComputedStyle(document.querySelector("."+header_class)).getPropertyValue("color");

                // let's replace the text with a link
                // keep same colour as before, but underline it to make it noticeable for user
                var newText = "<a href=\""+link+"\" style=\"text-decoration:underline;color:"+color+"\">"+pyImportPart+"</a>";
                cell.innerHTML = cell.innerHTML.replace(pyImportPart, newText);
            }
        }
    }
}