« get me outta code hell

Merge branch 'tf-parser' into staging - hsmusic-wiki - HSMusic - static wiki software cataloguing collaborative creation
about summary refs log tree commit diff
diff options
context:
space:
mode:
author(quasar) nebula <towerofnix@gmail.com>2021-05-07 15:31:48 -0300
committer(quasar) nebula <towerofnix@gmail.com>2021-05-07 15:31:48 -0300
commitcfb940ba021b1974c27b07dd0008b3bf065dad1e (patch)
tree8516adc329921f7ce1952df3d08e367f20400f98
parent5522bb97469e50f7762259d310d9db246ef8ce4a (diff)
parent405069fdd35849d4fede8183b14918e060d7b78c (diff)
Merge branch 'tf-parser' into staging
-rwxr-xr-xsrc/upd8.js439
-rw-r--r--src/util/sugar.js8
2 files changed, 425 insertions, 22 deletions
diff --git a/src/upd8.js b/src/upd8.js
index 26479119..3c157d34 100755
--- a/src/upd8.js
+++ b/src/upd8.js
@@ -141,6 +141,7 @@ import {
 
 import {
     call,
+    escapeRegex,
     filterEmptyLines,
     mapInPlace,
     queue,
@@ -794,8 +795,8 @@ const replacerSpec = {
     'flash': {
         search: 'flash',
         link: 'flash',
-        transformName(name, search, offset, text) {
-            const nextCharacter = text[offset + search.length];
+        transformName(name, node, input) {
+            const nextCharacter = input[node.iEnd];
             const lastCharacter = name[name.length - 1];
             if (
                 ![' ', '\n', '<'].includes(nextCharacter) &&
@@ -847,6 +848,11 @@ const replacerSpec = {
         search: 'staticPage',
         link: 'staticPage'
     },
+    'string': {
+        search: null,
+        value: ref => ref,
+        html: (ref, {strings, args}) => strings(ref, args)
+    },
     'tag': {
         search: 'tag',
         link: 'tag'
@@ -871,14 +877,353 @@ const replacerSpec = {
     }
     if (error) process.exit();
 
-    const categoryPart = Object.keys(replacerSpec).join('|');
-    transformInline.regexp = new RegExp(String.raw`(?<!\\)\[\[((${categoryPart}):)?(.+?)((?<! )#.+?)?(\|(.+?))?\]\]`, 'g');
+    // Syntax literals.
+    const tagBeginning = '[[';
+    const tagEnding = ']]';
+    const tagReplacerValue = ':';
+    const tagHash = '#';
+    const tagArgument = '*';
+    const tagArgumentValue = '=';
+    const tagLabel = '|';
+
+    const noPrecedingWhitespace = '(?<!\\s)';
+
+    const R_tagBeginning =
+        escapeRegex(tagBeginning);
+
+    const R_tagEnding =
+        escapeRegex(tagEnding);
+
+    const R_tagReplacerValue =
+        noPrecedingWhitespace +
+        escapeRegex(tagReplacerValue);
+
+    const R_tagHash =
+        noPrecedingWhitespace +
+        escapeRegex(tagHash);
+
+    const R_tagArgument =
+        escapeRegex(tagArgument);
+
+    const R_tagArgumentValue =
+        escapeRegex(tagArgumentValue);
+
+    const R_tagLabel =
+        escapeRegex(tagLabel);
+
+    const regexpCache = {};
+
+    const makeError = (i, message) => ({i, type: 'error', data: {message}});
+    const endOfInput = (i, comment) => makeError(i, `Unexpected end of input (${comment}).`);
+
+    // These are 8asically stored on the glo8al scope, which might seem odd
+    // for a recursive function, 8ut the values are only ever used immediately
+    // after they're set.
+    let stopped,
+        stop_iMatch,
+        stop_iParse,
+        stop_literal;
+
+    const parseOneTextNode = function(input, i, stopAt) {
+        return parseNodes(input, i, stopAt, true)[0];
+    };
+
+    const parseNodes = function(input, i, stopAt, textOnly) {
+        let nodes = [];
+        let escapeNext = false;
+        let string = '';
+        let iString = 0;
+
+        stopped = false;
+
+        const pushTextNode = (isLast) => {
+            string = input.slice(iString, i);
+
+            // If this is the last text node 8efore stopping (at a stopAt match
+            // or the end of the input), trim off whitespace at the end.
+            if (isLast) {
+                string = string.trimEnd();
+            }
+
+            if (string.length) {
+                nodes.push({i: iString, iEnd: i, type: 'text', data: string});
+                string = '';
+            }
+        };
+
+        const literalsToMatch = stopAt ? stopAt.concat([R_tagBeginning]) : [R_tagBeginning];
+
+        // The 8ackslash stuff here is to only match an even (or zero) num8er
+        // of sequential 'slashes. Even amounts always cancel out! Odd amounts
+        // don't, which would mean the following literal is 8eing escaped and
+        // should 8e counted only as part of the current string/text.
+        //
+        // Inspired 8y this: https://stackoverflow.com/a/41470813
+        const regexpSource = `(?<!\\\\)(?:\\\\{2})*(${literalsToMatch.join('|')})`;
+
+        // There are 8asically only a few regular expressions we'll ever use,
+        // 8ut it's a pain to hard-code them all, so we dynamically gener8te
+        // and cache them for reuse instead.
+        let regexp;
+        if (regexpCache.hasOwnProperty(regexpSource)) {
+            regexp = regexpCache[regexpSource];
+        } else {
+            regexp = new RegExp(regexpSource);
+            regexpCache[regexpSource] = regexp;
+        }
+
+        // Skip whitespace at the start of parsing. This is run every time
+        // parseNodes is called (and thus parseOneTextNode too), so spaces
+        // at the start of syntax elements will always 8e skipped. We don't
+        // skip whitespace that shows up inside content (i.e. once we start
+        // parsing below), though!
+        const whitespaceOffset = input.slice(i).search(/[^\s]/);
+
+        // If the string is all whitespace, that's just zero content, so
+        // return the empty nodes array.
+        if (whitespaceOffset === -1) {
+            return nodes;
+        }
+
+        i += whitespaceOffset;
+
+        while (i < input.length) {
+            const match = input.slice(i).match(regexp);
+
+            if (!match) {
+                iString = i;
+                i = input.length;
+                pushTextNode(true);
+                break;
+            }
+
+            const closestMatch = match[0];
+            const closestMatchIndex = i + match.index;
+
+            if (textOnly && closestMatch === tagBeginning)
+                throw makeError(i, `Unexpected [[tag]] - expected only text here.`);
+
+            const stopHere = (closestMatch !== tagBeginning);
+
+            iString = i;
+            i = closestMatchIndex;
+            pushTextNode(stopHere);
+
+            i += closestMatch.length;
+
+            if (stopHere) {
+                stopped = true;
+                stop_iMatch = closestMatchIndex;
+                stop_iParse = i;
+                stop_literal = closestMatch;
+                break;
+            }
+
+            if (closestMatch === tagBeginning) {
+                const iTag = closestMatchIndex;
+
+                let N;
+
+                // Replacer key (or value)
+
+                N = parseOneTextNode(input, i, [R_tagReplacerValue, R_tagHash, R_tagArgument, R_tagLabel, R_tagEnding]);
+
+                if (!stopped) throw endOfInput(i, `reading replacer key`);
+
+                if (!N) {
+                    switch (stop_literal) {
+                        case tagReplacerValue:
+                        case tagArgument:
+                            throw makeError(i, `Expected text (replacer key).`);
+                        case tagLabel:
+                        case tagHash:
+                        case tagEnding:
+                            throw makeError(i, `Expected text (replacer key/value).`);
+                    }
+                }
+
+                const replacerFirst = N;
+                i = stop_iParse;
+
+                // Replacer value (if explicit)
+
+                let replacerSecond;
+
+                if (stop_literal === tagReplacerValue) {
+                    N = parseNodes(input, i, [R_tagHash, R_tagArgument, R_tagLabel, R_tagEnding]);
+
+                    if (!stopped) throw endOfInput(i, `reading replacer value`);
+                    if (!N.length) throw makeError(i, `Expected content (replacer value).`);
+
+                    replacerSecond = N;
+                    i = stop_iParse
+                }
+
+                // Assign first & second to replacer key/value
+
+                let replacerKey,
+                    replacerValue;
+
+                // Value is an array of nodes, 8ut key is just one (or null).
+                // So if we use replacerFirst as the value, we need to stick
+                // it in an array (on its own).
+                if (replacerSecond) {
+                    replacerKey = replacerFirst;
+                    replacerValue = replacerSecond;
+                } else {
+                    replacerKey = null;
+                    replacerValue = [replacerFirst];
+                }
+
+                // Hash
+
+                let hash;
+
+                if (stop_literal === tagHash) {
+                    N = parseNodes(input, i, [R_tagArgument, R_tagLabel, R_tagEnding]);
+
+                    if (!stopped) throw endOfInput(i, `reading hash`);
+
+                    if (!N)
+                        throw makeError(i, `Expected content (hash).`);
+
+                    hash = N;
+                    i = stop_iParse;
+                }
+
+                // Arguments
+
+                const args = [];
+
+                while (stop_literal === tagArgument) {
+                    N = parseOneTextNode(input, i, [R_tagArgumentValue, R_tagArgument, R_tagLabel, R_tagEnding]);
+
+                    if (!stopped) throw endOfInput(i, `reading argument key`);
+
+                    if (stop_literal !== tagArgumentValue)
+                        throw makeError(i, `Expected ${tagArgumentValue.literal} (tag argument).`);
+
+                    if (!N)
+                        throw makeError(i, `Expected text (argument key).`);
+
+                    const key = N;
+                    i = stop_iParse;
+
+                    N = parseNodes(input, i, [R_tagArgument, R_tagLabel, R_tagEnding]);
+
+                    if (!stopped) throw endOfInput(i, `reading argument value`);
+                    if (!N.length) throw makeError(i, `Expected content (argument value).`);
+
+                    const value = N;
+                    i = stop_iParse;
+
+                    args.push({key, value});
+                }
+
+                let label;
+
+                if (stop_literal === tagLabel) {
+                    N = parseOneTextNode(input, i, [R_tagEnding]);
+
+                    if (!stopped) throw endOfInput(i, `reading label`);
+                    if (!N) throw makeError(i, `Expected text (label).`);
+
+                    label = N;
+                    i = stop_iParse;
+                }
+
+                nodes.push({i: iTag, iEnd: i, type: 'tag', data: {replacerKey, replacerValue, hash, args, label}});
+
+                continue;
+            }
+        }
+
+        return nodes;
+    };
+
+    transformInline.parse = function(input) {
+        try {
+            return parseNodes(input, 0);
+        } catch (errorNode) {
+            if (errorNode.type !== 'error') {
+                throw errorNode;
+            }
+
+            const { i, data: { message } } = errorNode;
+
+            let lineStart = input.slice(0, i).lastIndexOf('\n');
+            if (lineStart >= 0) {
+                lineStart += 1;
+            } else {
+                lineStart = 0;
+            }
+
+            let lineEnd = input.slice(i).indexOf('\n');
+            if (lineEnd >= 0) {
+                lineEnd += i;
+            } else {
+                lineEnd = input.length;
+            }
+
+            const line = input.slice(lineStart, lineEnd);
+
+            const cursor = i - lineStart;
+
+            throw new SyntaxError(fixWS`
+                Parse error (at pos ${i}): ${message}
+                ${line}
+                ${'-'.repeat(cursor) + '^'}
+            `);
+        }
+    };
+}
+
+/*
+{
+    const show = input => process.stdout.write(`-- ${input}\n` + util.inspect(
+        transformInline.parse(input),
+        {
+            depth: null,
+            colors: true
+        }
+    ) + '\n\n');
+
+    show(`[[album:are-you-lost|Cristata's new album]]`);
+    show(`[[string:content.donate.patreonLine*link=[[external:https://www.patreon.com/qznebula|Patreon]]]]`);
+}
+
+{
+    const test = input => {
+        let n = 0;
+        const s = 5;
+        const start = Date.now();
+        const end = start + s * 1000;
+        while (Date.now() < end) {
+            transformInline.parse(input);
+            n++;
+        }
+        console.log(`Ran ${Math.round(n / s)} times/sec.`);
+    }
+
+    test(fixWS`
+        [[string:content.donate.patreonLine*link=[[external:https://www.patreon.com/qznebula|Patreon]]]]
+        Hello, world! Wow [[album:the-beans-zone]] is some cool stuff.
+    `);
+    process.exit();
 }
+*/
+
+{
+    const evaluateTag = function(node, opts) {
+        const { input, strings, to } = opts;
+
+        const source = input.slice(node.i, node.iEnd);
 
-function transformInline(text, {strings, to}) {
-    return text.replace(transformInline.regexp, (match, _1, category, ref, hash, _2, enteredName, offset) => {
-        if (!category) {
-            category = 'track';
+        const replacerKey = node.data.replacerKey?.data || 'track';
+
+        if (!replacerSpec[replacerKey]) {
+            logWarn`The link ${source} has an invalid replacer key!`;
+            return source;
         }
 
         const {
@@ -887,41 +1232,91 @@ function transformInline(text, {strings, to}) {
             value: valueFn,
             html: htmlFn,
             transformName
-        } = replacerSpec[category];
+        } = replacerSpec[replacerKey];
+
+        const replacerValue = transformNodes(node.data.replacerValue, opts);
 
         const value = (
-            valueFn ? valueFn(ref) :
-            searchKey ? search[searchKey](ref) :
+            valueFn ? valueFn(replacerValue) :
+            searchKey ? search[searchKey](replacerValue) :
             {
-                directory: ref.replace(category + ':', ''),
+                directory: replacerValue,
                 name: null
             });
 
         if (!value) {
-            logWarn`The link ${match} does not match anything!`;
-            return match;
+            logWarn`The link ${search} does not match anything!`;
+            return search;
         }
 
-        const label = (enteredName
-            || transformName && transformName(value.name, match, offset, text)
+        const enteredLabel = node.data.label && transformNode(node.data.label, opts);
+
+        const label = (enteredLabel
+            || transformName && transformName(value.name, node, input)
             || value.name);
 
         if (!valueFn && !label) {
-            logWarn`The link ${match} requires a label be entered!`;
-            return match;
+            logWarn`The link ${search} requires a label be entered!`;
+            return search;
         }
 
+        const hash = node.data.hash && transformNodes(node.data.hash, opts);
+
+        const args = node.data.args && Object.fromEntries(node.data.args.map(
+            ({ key, value }) => [
+                transformNode(key, opts),
+                transformNodes(value, opts)
+            ]));
+
         const fn = (htmlFn
             ? htmlFn
             : strings.link[linkKey]);
 
         try {
-            return fn(value, {text: label, hash, strings, to});
+            return fn(value, {text: label, hash, args, strings, to});
         } catch (error) {
-            logError`The link ${match} failed to be processed: ${error}`;
-            return match;
+            logError`The link ${source} failed to be processed: ${error}`;
+            return source;
+        }
+    };
+
+    const transformNode = function(node, opts) {
+        if (!node) {
+            throw new Error('Expected a node!');
+        }
+
+        if (Array.isArray(node)) {
+            throw new Error('Got an array - use transformNodes here!');
+        }
+
+        switch (node.type) {
+            case 'text':
+                return node.data;
+            case 'tag':
+                return evaluateTag(node, opts);
+            default:
+                throw new Error(`Unknown node type ${node.type}`);
         }
-    }).replaceAll(String.raw`\[[`, '[[');
+    };
+
+    const transformNodes = function(nodes, opts) {
+        if (!nodes || !Array.isArray(nodes)) {
+            throw new Error(`Expected an array of nodes! Got: ${nodes}`);
+        }
+
+        return nodes.map(node => transformNode(node, opts)).join('');
+    };
+
+    Object.assign(transformInline, {
+        evaluateTag,
+        transformNode,
+        transformNodes
+    });
+}
+
+function transformInline(input, {strings, to}) {
+    const nodes = transformInline.parse(input);
+    return transformInline.transformNodes(nodes, {strings, to, input});
 }
 
 function parseAttributes(string, {to}) {
diff --git a/src/util/sugar.js b/src/util/sugar.js
index 9970bad1..c24c617c 100644
--- a/src/util/sugar.js
+++ b/src/util/sugar.js
@@ -68,3 +68,11 @@ export function queue(array, max = 50) {
 export function delay(ms) {
     return new Promise(res => setTimeout(res, ms));
 }
+
+// Stolen from here: https://stackoverflow.com/a/3561711
+//
+// There's a proposal for a native JS function like this, 8ut it's not even
+// past stage 1 yet: https://github.com/tc39/proposal-regex-escaping
+export function escapeRegex(string) {
+    return string.replace(/[-\/\\^$*+?.()|[\]{}]/g, '\\$&');
+}