forked from angular/angular
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
fix(migrations): Switches to multiple passes to fix several reported …
…bugs (angular#52592) Rather than migrate all in one pass, this now migrates in a separate pass per control flow item plus one for templates at the end. This resolves issues with multiple control flow items on a single element as well as making sure ng-templates are fully migrated before being moved to new locations. fixes: angular#52518 fixes: angular#52516 fixes: angular#52513 PR Close angular#52592
- Loading branch information
1 parent
682bd39
commit b131d61
Showing
7 changed files
with
666 additions
and
402 deletions.
There are no files selected for viewing
181 changes: 181 additions & 0 deletions
181
packages/core/schematics/ng-generate/control-flow-migration/fors.ts
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
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,181 @@ | ||
/** | ||
* @license | ||
* Copyright Google LLC All Rights Reserved. | ||
* | ||
* Use of this source code is governed by an MIT-style license that can be | ||
* found in the LICENSE file at https://angular.io/license | ||
*/ | ||
|
||
import {visitAll} from '@angular/compiler'; | ||
|
||
import {ElementCollector, ElementToMigrate, MigrateError, Result} from './types'; | ||
import {calculateNesting, getMainBlock, getOriginals, hasLineBreaks, parseTemplate, reduceNestingOffset} from './util'; | ||
|
||
export const ngfor = '*ngFor'; | ||
export const commaSeparatedSyntax = new Map([ | ||
['(', ')'], | ||
['{', '}'], | ||
['[', ']'], | ||
]); | ||
export const stringPairs = new Map([ | ||
[`"`, `"`], | ||
[`'`, `'`], | ||
]); | ||
|
||
/** | ||
* Replaces structural directive ngFor instances with new for. | ||
* Returns null if the migration failed (e.g. there was a syntax error). | ||
*/ | ||
export function migrateFor(template: string): {migrated: string, errors: MigrateError[]} { | ||
let errors: MigrateError[] = []; | ||
let parsed = parseTemplate(template); | ||
if (parsed === null) { | ||
return {migrated: template, errors}; | ||
} | ||
|
||
let result = template; | ||
const visitor = new ElementCollector([ngfor]); | ||
visitAll(visitor, parsed.rootNodes); | ||
calculateNesting(visitor, hasLineBreaks(template)); | ||
|
||
// this tracks the character shift from different lengths of blocks from | ||
// the prior directives so as to adjust for nested block replacement during | ||
// migration. Each block calculates length differences and passes that offset | ||
// to the next migrating block to adjust character offsets properly. | ||
let offset = 0; | ||
let nestLevel = -1; | ||
let postOffsets: number[] = []; | ||
for (const el of visitor.elements) { | ||
let migrateResult: Result = {tmpl: result, offsets: {pre: 0, post: 0}}; | ||
// applies the post offsets after closing | ||
offset = reduceNestingOffset(el, nestLevel, offset, postOffsets); | ||
|
||
try { | ||
migrateResult = migrateNgFor(el, result, offset); | ||
} catch (error: unknown) { | ||
errors.push({type: ngfor, error}); | ||
} | ||
|
||
result = migrateResult.tmpl; | ||
offset += migrateResult.offsets.pre; | ||
postOffsets.push(migrateResult.offsets.post); | ||
nestLevel = el.nestCount; | ||
} | ||
|
||
return {migrated: result, errors}; | ||
} | ||
|
||
|
||
function migrateNgFor(etm: ElementToMigrate, tmpl: string, offset: number): Result { | ||
const aliasWithEqualRegexp = /=\s*(count|index|first|last|even|odd)/gm; | ||
const aliasWithAsRegexp = /(count|index|first|last|even|odd)\s+as/gm; | ||
const aliases = []; | ||
const lbString = etm.hasLineBreaks ? '\n' : ''; | ||
const lbSpaces = etm.hasLineBreaks ? `\n ` : ''; | ||
const parts = getNgForParts(etm.attr.value); | ||
|
||
const originals = getOriginals(etm, tmpl, offset); | ||
|
||
// first portion should always be the loop definition prefixed with `let` | ||
const condition = parts[0].replace('let ', ''); | ||
const loopVar = condition.split(' of ')[0]; | ||
let trackBy = loopVar; | ||
let aliasedIndex: string|null = null; | ||
for (let i = 1; i < parts.length; i++) { | ||
const part = parts[i].trim(); | ||
|
||
if (part.startsWith('trackBy:')) { | ||
// build trackby value | ||
const trackByFn = part.replace('trackBy:', '').trim(); | ||
trackBy = `${trackByFn}($index, ${loopVar})`; | ||
} | ||
// aliases | ||
// declared with `let myIndex = index` | ||
if (part.match(aliasWithEqualRegexp)) { | ||
// 'let myIndex = index' -> ['let myIndex', 'index'] | ||
const aliasParts = part.split('='); | ||
// -> 'let myIndex = $index' | ||
aliases.push(` ${aliasParts[0].trim()} = $${aliasParts[1].trim()}`); | ||
// if the aliased variable is the index, then we store it | ||
if (aliasParts[1].trim() === 'index') { | ||
// 'let myIndex' -> 'myIndex' | ||
aliasedIndex = aliasParts[0].replace('let', '').trim(); | ||
} | ||
} | ||
// declared with `index as myIndex` | ||
if (part.match(aliasWithAsRegexp)) { | ||
// 'index as myIndex' -> ['index', 'myIndex'] | ||
const aliasParts = part.split(/\s+as\s+/); | ||
// -> 'let myIndex = $index' | ||
aliases.push(` let ${aliasParts[1].trim()} = $${aliasParts[0].trim()}`); | ||
// if the aliased variable is the index, then we store it | ||
if (aliasParts[0].trim() === 'index') { | ||
aliasedIndex = aliasParts[1].trim(); | ||
} | ||
} | ||
} | ||
// if an alias has been defined for the index, then the trackBy function must use it | ||
if (aliasedIndex !== null && trackBy !== loopVar) { | ||
// byId($index, user) -> byId(i, user) | ||
trackBy = trackBy.replace('$index', aliasedIndex); | ||
} | ||
|
||
const aliasStr = (aliases.length > 0) ? `;${aliases.join(';')}` : ''; | ||
|
||
const {start, middle, end} = getMainBlock(etm, tmpl, offset); | ||
const startBlock = `@for (${condition}; track ${trackBy}${aliasStr}) {${lbSpaces}${start}`; | ||
|
||
const endBlock = `${end}${lbString}}`; | ||
const forBlock = startBlock + middle + endBlock; | ||
|
||
const updatedTmpl = tmpl.slice(0, etm.start(offset)) + forBlock + tmpl.slice(etm.end(offset)); | ||
|
||
const pre = originals.start.length - startBlock.length; | ||
const post = originals.end.length - endBlock.length; | ||
|
||
return {tmpl: updatedTmpl, offsets: {pre, post}}; | ||
} | ||
|
||
function getNgForParts(expression: string): string[] { | ||
const parts: string[] = []; | ||
const commaSeparatedStack: string[] = []; | ||
const stringStack: string[] = []; | ||
let current = ''; | ||
|
||
for (let i = 0; i < expression.length; i++) { | ||
const char = expression[i]; | ||
const isInString = stringStack.length === 0; | ||
const isInCommaSeparated = commaSeparatedStack.length === 0; | ||
|
||
// Any semicolon is a delimiter, as well as any comma outside | ||
// of comma-separated syntax, as long as they're outside of a string. | ||
if (isInString && current.length > 0 && | ||
(char === ';' || (char === ',' && isInCommaSeparated))) { | ||
parts.push(current); | ||
current = ''; | ||
continue; | ||
} | ||
|
||
if (stringPairs.has(char)) { | ||
stringStack.push(stringPairs.get(char)!); | ||
} else if (stringStack.length > 0 && stringStack[stringStack.length - 1] === char) { | ||
stringStack.pop(); | ||
} | ||
|
||
if (commaSeparatedSyntax.has(char)) { | ||
commaSeparatedStack.push(commaSeparatedSyntax.get(char)!); | ||
} else if ( | ||
commaSeparatedStack.length > 0 && | ||
commaSeparatedStack[commaSeparatedStack.length - 1] === char) { | ||
commaSeparatedStack.pop(); | ||
} | ||
|
||
current += char; | ||
} | ||
|
||
if (current.length > 0) { | ||
parts.push(current); | ||
} | ||
|
||
return parts; | ||
} |
156 changes: 156 additions & 0 deletions
156
packages/core/schematics/ng-generate/control-flow-migration/ifs.ts
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
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,156 @@ | ||
/** | ||
* @license | ||
* Copyright Google LLC All Rights Reserved. | ||
* | ||
* Use of this source code is governed by an MIT-style license that can be | ||
* found in the LICENSE file at https://angular.io/license | ||
*/ | ||
|
||
import {visitAll} from '@angular/compiler'; | ||
|
||
import {ElementCollector, ElementToMigrate, MigrateError, Result} from './types'; | ||
import {calculateNesting, getMainBlock, getOriginals, hasLineBreaks, parseTemplate, reduceNestingOffset} from './util'; | ||
|
||
export const ngif = '*ngIf'; | ||
export const boundngif = '[ngIf]'; | ||
export const nakedngif = 'ngIf'; | ||
|
||
export const ifs = [ | ||
ngif, | ||
nakedngif, | ||
boundngif, | ||
]; | ||
|
||
/** | ||
* Replaces structural directive ngif instances with new if. | ||
* Returns null if the migration failed (e.g. there was a syntax error). | ||
*/ | ||
export function migrateIf(template: string): {migrated: string, errors: MigrateError[]} { | ||
let errors: MigrateError[] = []; | ||
let parsed = parseTemplate(template); | ||
if (parsed === null) { | ||
return {migrated: template, errors}; | ||
} | ||
|
||
let result = template; | ||
const visitor = new ElementCollector(ifs); | ||
visitAll(visitor, parsed.rootNodes); | ||
calculateNesting(visitor, hasLineBreaks(template)); | ||
|
||
// this tracks the character shift from different lengths of blocks from | ||
// the prior directives so as to adjust for nested block replacement during | ||
// migration. Each block calculates length differences and passes that offset | ||
// to the next migrating block to adjust character offsets properly. | ||
let offset = 0; | ||
let nestLevel = -1; | ||
let postOffsets: number[] = []; | ||
for (const el of visitor.elements) { | ||
let migrateResult: Result = {tmpl: result, offsets: {pre: 0, post: 0}}; | ||
// applies the post offsets after closing | ||
offset = reduceNestingOffset(el, nestLevel, offset, postOffsets); | ||
|
||
try { | ||
migrateResult = migrateNgIf(el, result, offset); | ||
} catch (error: unknown) { | ||
errors.push({type: ngif, error}); | ||
} | ||
|
||
result = migrateResult.tmpl; | ||
offset += migrateResult.offsets.pre; | ||
postOffsets.push(migrateResult.offsets.post); | ||
nestLevel = el.nestCount; | ||
} | ||
|
||
return {migrated: result, errors}; | ||
} | ||
|
||
export function migrateNgIf(etm: ElementToMigrate, tmpl: string, offset: number): Result { | ||
const matchThen = etm.attr.value.match(/;\s*then/gm); | ||
const matchElse = etm.attr.value.match(/;\s*else/gm); | ||
|
||
if (matchThen && matchThen.length > 0) { | ||
return buildIfThenElseBlock(etm, tmpl, matchThen[0], matchElse![0], offset); | ||
} else if (matchElse && matchElse.length > 0) { | ||
// just else | ||
return buildIfElseBlock(etm, tmpl, matchElse[0], offset); | ||
} | ||
|
||
return buildIfBlock(etm, tmpl, offset); | ||
} | ||
|
||
function buildIfBlock(etm: ElementToMigrate, tmpl: string, offset: number): Result { | ||
// includes the mandatory semicolon before as | ||
const lbString = etm.hasLineBreaks ? '\n' : ''; | ||
const condition = etm.attr.value.replace(' as ', '; as '); | ||
|
||
const originals = getOriginals(etm, tmpl, offset); | ||
|
||
const {start, middle, end} = getMainBlock(etm, tmpl, offset); | ||
const startBlock = `@if (${condition}) {${lbString}${start}`; | ||
const endBlock = `${end}${lbString}}`; | ||
|
||
const ifBlock = startBlock + middle + endBlock; | ||
const updatedTmpl = tmpl.slice(0, etm.start(offset)) + ifBlock + tmpl.slice(etm.end(offset)); | ||
|
||
// this should be the difference between the starting element up to the start of the closing | ||
// element and the mainblock sans } | ||
const pre = originals.start.length - startBlock.length; | ||
const post = originals.end.length - endBlock.length; | ||
|
||
return {tmpl: updatedTmpl, offsets: {pre, post}}; | ||
} | ||
|
||
function buildIfElseBlock( | ||
etm: ElementToMigrate, tmpl: string, elseString: string, offset: number): Result { | ||
// includes the mandatory semicolon before as | ||
const lbString = etm.hasLineBreaks ? '\n' : ''; | ||
const condition = etm.getCondition(elseString).replace(' as ', '; as '); | ||
|
||
const originals = getOriginals(etm, tmpl, offset); | ||
|
||
const elsePlaceholder = `#${etm.getTemplateName(elseString)}|`; | ||
const {start, middle, end} = getMainBlock(etm, tmpl, offset); | ||
const startBlock = `@if (${condition}) {${lbString}${start}`; | ||
|
||
const elseBlock = `${end}${lbString}} @else {${lbString}`; | ||
|
||
const postBlock = elseBlock + elsePlaceholder + `${lbString}}`; | ||
const ifElseBlock = startBlock + middle + postBlock; | ||
|
||
const tmplStart = tmpl.slice(0, etm.start(offset)); | ||
const tmplEnd = tmpl.slice(etm.end(offset)); | ||
const updatedTmpl = tmplStart + ifElseBlock + tmplEnd; | ||
|
||
const pre = originals.start.length - startBlock.length; | ||
const post = originals.end.length - postBlock.length; | ||
|
||
return {tmpl: updatedTmpl, offsets: {pre, post}}; | ||
} | ||
|
||
function buildIfThenElseBlock( | ||
etm: ElementToMigrate, tmpl: string, thenString: string, elseString: string, | ||
offset: number): Result { | ||
const condition = etm.getCondition(thenString).replace(' as ', '; as '); | ||
const lbString = etm.hasLineBreaks ? '\n' : ''; | ||
|
||
const originals = getOriginals(etm, tmpl, offset); | ||
|
||
const startBlock = `@if (${condition}) {${lbString}`; | ||
const elseBlock = `${lbString}} @else {${lbString}`; | ||
|
||
const thenPlaceholder = `#${etm.getTemplateName(thenString, elseString)}|`; | ||
const elsePlaceholder = `#${etm.getTemplateName(elseString)}|`; | ||
|
||
const postBlock = thenPlaceholder + elseBlock + elsePlaceholder + `${lbString}}`; | ||
const ifThenElseBlock = startBlock + postBlock; | ||
|
||
const tmplStart = tmpl.slice(0, etm.start(offset)); | ||
const tmplEnd = tmpl.slice(etm.end(offset)); | ||
|
||
const updatedTmpl = tmplStart + ifThenElseBlock + tmplEnd; | ||
|
||
const pre = originals.start.length - startBlock.length; | ||
const post = originals.end.length - postBlock.length; | ||
|
||
return {tmpl: updatedTmpl, offsets: {pre, post}}; | ||
} |
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
Oops, something went wrong.