houdini 0.17.9 → 0.17.10

package/src/codegen/generators/artifacts/fieldKey.ts

@@ -1,60 +0,0 @@
-import * as graphql from 'graphql'
-
-import { Config } from '../../../lib'
-
-// we need to generate a static key that we can use to index this field in the cache.
-// this needs to be a unique hash driven by the field's attribute and arguments
-// returns the key for a specific field
-export default function fieldKey(config: Config, field: graphql.FieldNode): string {
-	// we're going to hash a field by creating a json object and adding it
-	// to the attribute name
-	const attributeName = field.alias?.value || field.name.value
-
-	// field might not have a location so print and re-parse before we look at serialized values
-	const printed = graphql.print(field)
-	const secondParse = (
-		graphql.parse(`{${printed}}`).definitions[0] as graphql.OperationDefinitionNode
-	).selectionSet.selections[0] as graphql.FieldNode
-
-	// if the field is paginated, we need to strip away any args
-	const paginated = !!field.directives?.find(
-		(directive) => directive.name.value === config.paginateDirective
-	)
-	const paginationArgs = ['first', 'after', 'last', 'before', 'limit', 'offset']
-
-	const argObj = (secondParse.arguments || []).reduce<{ [key: string]: string }>((acc, arg) => {
-		// the query already contains a serialized version of the argument so just pull it out of the
-		// document string
-		const start = arg.value.loc?.start
-		const end = arg.value.loc?.end
-
-		// if the field is paginated, ignore the pagination args in the key
-		if (paginated && paginationArgs.includes(arg.name.value)) {
-			return acc
-		}
-
-		// if the argument is not in the query, life doesn't make sense
-		if (!start || !end) {
-			return acc
-		}
-
-		return {
-			...acc,
-			[arg.name.value]: printed.substring(start - 1, end - 1),
-		}
-	}, {})
-
-	let key =
-		Object.values(argObj).length > 0
-			? `${attributeName}(${Object.entries(argObj)
-					.map((entries) => entries.join(': '))
-					.join(', ')})`
-			: attributeName
-
-	// if the field is paginated, key it differently so other documents can ask for the non paginated value without conflict
-	if (paginated) {
-		key = key + '::paginated'
-	}
-
-	return key
-}

package/src/codegen/generators/artifacts/index.ts

@@ -1,330 +0,0 @@
-import * as graphql from 'graphql'
-import * as recast from 'recast'
-
-import {
-	Config,
-	getRootType,
-	hashDocument,
-	HoudiniError,
-	parentTypeFromAncestors,
-	fs,
-	CollectedGraphQLDocument,
-	cleanupFiles,
-} from '../../../lib'
-import { ArtifactKind } from '../../../lib'
-import { moduleExport } from '../../utils'
-import writeIndexFile from './indexFile'
-import { inputObject } from './inputs'
-import { operationsByPath, FilterMap } from './operations'
-import selection from './selection'
-import { serializeValue } from './utils'
-
-const AST = recast.types.builders
-
-// the artifact generator creates files in the runtime directory for each
-// document containing meta data that the preprocessor might use
-export default function artifactGenerator(stats: {
-	total: string[]
-	new: string[]
-	changed: string[]
-	deleted: string[]
-}) {
-	return async function (config: Config, docs: CollectedGraphQLDocument[]) {
-		// put together the type information for the filter for every list
-		const filterTypes: FilterMap = {}
-
-		for (const doc of docs) {
-			graphql.visit(doc.document, {
-				// look for any field marked with a list
-				Directive(node, _, __, ___, ancestors) {
-					// we only care about lists
-					if (node.name.value !== config.listDirective) {
-						return
-					}
-
-					// get the name of the list
-					const nameArg = node.arguments?.find(
-						(arg) => arg.name.value === config.listNameArg
-					)
-					if (!nameArg || nameArg.value.kind !== 'StringValue') {
-						throw new HoudiniError({
-							filepath: doc.filename,
-							message: 'could not find name arg in list directive',
-						})
-					}
-					const listName = nameArg.value.value
-
-					// look up the actual field in the ancestor list so we can get type info
-					let field = ancestors[ancestors.length - 1] as graphql.FieldNode
-					let i = 1
-					while (Array.isArray(field)) {
-						i++
-						field = ancestors[ancestors.length - i] as graphql.FieldNode
-					}
-					if (field.kind !== 'Field') {
-						return
-					}
-
-					// look up the parent's type so we can ask about the field marked as a list
-					const parentType = parentTypeFromAncestors(config.schema, doc.filename, [
-						...ancestors.slice(0, -1),
-					]) as graphql.GraphQLObjectType
-					const parentField = parentType.getFields()[field.name.value]
-					if (!parentField) {
-						throw new HoudiniError({
-							filepath: doc.filename,
-							message: 'Could not find field information when computing filters',
-						})
-					}
-					const fieldType = getRootType(parentField.type).toString()
-
-					// look at every arg on the list to figure out the valid filters
-					filterTypes[listName] = parentField.args.reduce((prev, arg) => {
-						return {
-							...prev,
-							[arg.name]: getRootType(arg.type).toString(),
-						}
-					}, {})
-
-					// the delete directive is an interesting one since there isn't a specific
-					// list. we need to use something that points to deleting an instance of
-					// the type as a key
-					filterTypes[`${fieldType}_delete`] = {
-						...filterTypes[`${fieldType}_delete`],
-						// every field with the list type adds to the delete filters
-						...filterTypes[listName],
-					}
-				},
-			})
-		}
-
-		const listOfArtifacts: string[] = []
-
-		// we have everything we need to generate the artifacts
-		await Promise.all(
-			[
-				// generate the index file
-				writeIndexFile(config, docs),
-			].concat(
-				// and an artifact for every document
-				docs.map(async (doc) => {
-					// pull out the info we need from the collected doc
-					const { document, name, generateArtifact } = doc
-
-					// if the document is generated, don't write it to disk - it's use is to provide definitions
-					// for the other transforms
-					if (!generateArtifact) {
-						return
-					}
-
-					// before we can print the document, we need to strip:
-					// 1. all references to internal directives
-					// 2. all variables only used by internal directives
-					const usedVariableNames = new Set<string>()
-					let documentWithoutInternalDirectives = graphql.visit(document, {
-						Directive(node) {
-							// if the directive is one of the internal ones, remove it
-							if (config.isInternalDirective(node)) {
-								return null
-							}
-						},
-
-						Variable(node, _key, parent) {
-							const variableIsBeingDefined =
-								parent &&
-								!(parent instanceof Array) &&
-								parent.kind === 'VariableDefinition'
-
-							if (!variableIsBeingDefined) {
-								usedVariableNames.add(node.name.value)
-							}
-						},
-					})
-					let documentWithoutExtraVariables = graphql.visit(
-						documentWithoutInternalDirectives,
-						{
-							VariableDefinition(variableDefinitionNode) {
-								const name = variableDefinitionNode.variable.name.value
-
-								if (!usedVariableNames.has(name)) {
-									return null
-								}
-							},
-						}
-					)
-					let rawString = graphql.print(documentWithoutExtraVariables)
-
-					// figure out the document kind
-					let docKind = doc.kind
-
-					// look for the operation
-					const operations = document.definitions.filter(
-						({ kind }) => kind === graphql.Kind.OPERATION_DEFINITION
-					) as graphql.OperationDefinitionNode[]
-					// there are no operations, so its a fragment
-					const fragments = document.definitions.filter(
-						({ kind }) => kind === graphql.Kind.FRAGMENT_DEFINITION
-					) as graphql.FragmentDefinitionNode[]
-
-					let rootType: string | undefined = ''
-					let selectionSet: graphql.SelectionSetNode
-
-					// if we are generating the artifact for an operation
-					if (docKind !== ArtifactKind.Fragment) {
-						// find the operation
-						const operation = operations[0]
-
-						if (operation.operation === 'query') {
-							rootType = config.schema.getQueryType()?.name
-						} else if (operation.operation === 'mutation') {
-							rootType = config.schema.getMutationType()?.name
-						} else if (operation.operation === 'subscription') {
-							rootType = config.schema.getSubscriptionType()?.name
-						}
-						if (!rootType) {
-							throw new HoudiniError({
-								filepath: doc.filename,
-								message:
-									'could not find root type for operation: ' +
-									operation.operation +
-									'. Maybe you need to re-run the introspection query?',
-							})
-						}
-
-						// use this selection set
-						selectionSet = operation.selectionSet
-					}
-					// we are looking at a fragment so use its selection set and type for the subscribe index
-					else {
-						// there are a lot of fragments added to a document. The fragment we care about
-						// is the one with the matching name
-						const matchingFragment = fragments.find(
-							(fragment) => fragment.name.value === name
-						)
-						if (!matchingFragment) {
-							throw new HoudiniError({
-								filepath: doc.filename,
-								message: `Fragment "${name}" doesn't exist in its own document?!`,
-							})
-						}
-						rootType = matchingFragment.typeCondition.name.value
-						selectionSet = matchingFragment.selectionSet
-					}
-
-					// if there are inputs to the operation
-					const inputs = operations[0]?.variableDefinitions
-
-					// generate a hash of the document that we can use to detect changes
-					// start building up the artifact
-					const artifact: Record<string, any> = {
-						name,
-						kind: docKind,
-						hash: hashDocument(rawString),
-						refetch: doc.refetch,
-						raw: rawString,
-						rootType,
-						selection: selection({
-							config,
-							filepath: doc.filename,
-							rootType,
-							selections: selectionSet.selections,
-							operations: operationsByPath(
-								config,
-								doc.filename,
-								operations[0],
-								filterTypes
-							),
-							// do not include used fragments if we are rendering the selection
-							// for a fragment document
-							includeFragments: docKind !== 'HoudiniFragment',
-							document: doc,
-						}),
-					}
-
-					// if the document has inputs describe their types in the artifact so we can
-					// marshal and unmarshal scalars
-					if (inputs && inputs.length > 0) {
-						artifact.input = inputObject(config, inputs)
-					}
-
-					// add the cache policy to query documents
-					if (docKind === 'HoudiniQuery') {
-						const cacheDirective = operations[0].directives?.find(
-							(directive) => directive.name.value === config.cacheDirective
-						)
-						if (cacheDirective) {
-							// look for arguments
-							const args: { [key: string]: graphql.ArgumentNode } =
-								cacheDirective.arguments?.reduce(
-									(acc, arg) => ({
-										...acc,
-										[arg.name.value]: arg,
-									}),
-									{}
-								) || {}
-
-							const policy = args[config.cachePolicyArg]
-							if (policy && policy.value.kind === 'EnumValue') {
-								artifact.policy = policy.value.value
-							} else {
-								artifact.policy = config.defaultCachePolicy
-							}
-
-							// if the user opted-in for partial data
-							const partial = args[config.cachePartialArg]
-							if (partial && partial.value.kind === 'BooleanValue') {
-								artifact.partial = partial.value.value
-							} else {
-								artifact.partial = config.defaultPartial
-							}
-						} else {
-							artifact.policy = config.defaultCachePolicy
-							artifact.partial = config.defaultPartial
-						}
-					}
-
-					// the artifact should be the default export of the file
-					const file = AST.program([
-						moduleExport(config, 'default', serializeValue(artifact)),
-						AST.expressionStatement(
-							AST.stringLiteral(`HoudiniHash=${hashDocument(doc.originalString)}`)
-						),
-					])
-
-					const artifactPath = config.artifactPath(document)
-
-					// don't count the document unless it's user-facing (ie, generates a store)
-					const countDocument = doc.generateStore
-
-					// check if the file exists (indicating a new document)
-					let existingArtifact = await fs.readFile(artifactPath)
-					if (existingArtifact === null) {
-						if (countDocument) {
-							stats.new.push(artifact.name)
-						}
-					}
-
-					// write the result to the artifact path we're configured to write to
-					await fs.writeFile(artifactPath, recast.print(file).code)
-					listOfArtifacts.push(config.documentName(document))
-
-					if (!countDocument) {
-						return
-					}
-
-					// check if the artifact exists
-					const match = existingArtifact && existingArtifact.match(/"HoudiniHash=(\w+)"/)
-					if (match && match[1] !== hashDocument(doc.originalString)) {
-						stats.changed.push(artifact.name)
-					}
-
-					// regardless of whether it was changed or not, we need to track the total list of artifacts
-					stats.total.push(artifact.name)
-				})
-			)
-		)
-
-		// cleanup files that are no more necessary!
-		stats.deleted = await cleanupFiles(config.artifactDirectory, listOfArtifacts)
-	}
-}

package/src/codegen/generators/artifacts/indexFile.ts

@@ -1,24 +0,0 @@
-import { Config, fs, CollectedGraphQLDocument, path } from '../../../lib'
-import { cjsIndexFilePreamble, exportDefaultFrom } from '../../utils'
-
-export default async function writeIndexFile(config: Config, docs: CollectedGraphQLDocument[]) {
-	const docsToGenerate = docs
-		.filter((doc) => doc.generateArtifact)
-		.sort((a, b) => a.name.localeCompare(b.name))
-
-	// we want to export every artifact from the index file.
-	let body =
-		config.module === 'esm'
-			? docsToGenerate.reduce(
-					(content, doc) =>
-						content + `\n export { default as ${doc.name}} from './${doc.name}'`,
-					''
-			  )
-			: docsToGenerate.reduce(
-					(content, doc) => content + `\n${exportDefaultFrom(`./${doc.name}`, doc.name)}`,
-					cjsIndexFilePreamble
-			  )
-
-	// write the result to the artifact path we're configured to write to
-	await fs.writeFile(path.join(config.artifactDirectory, 'index.js'), body)
-}

package/src/codegen/generators/artifacts/inputs.ts

@@ -1,81 +0,0 @@
-import * as graphql from 'graphql'
-import * as recast from 'recast'
-
-import { Config } from '../../../lib/config'
-import { InputObject } from '../../../runtime/lib/types'
-import { unwrapType } from '../../utils'
-
-const AST = recast.types.builders
-
-export function inputObject(
-	config: Config,
-	inputs: readonly graphql.VariableDefinitionNode[]
-): InputObject {
-	// make sure we don't define the same input type
-	const visitedTypes = new Set<string>()
-
-	// inputs can be recursive so we can't flatten the input type into a single object
-	const inputObj: InputObject = {
-		fields: inputs.reduce((fields, input) => {
-			// find the inner type
-			const { type } = unwrapType(config, input.type)
-
-			// embed the type in the input
-			return {
-				...fields,
-				[input.variable.name.value]: type.name,
-			}
-		}, {}),
-		types: {},
-	}
-
-	// walk through every type referenced and add it to the list
-	for (const input of inputs) {
-		walkInputs(config, visitedTypes, inputObj, input.type)
-	}
-
-	return inputObj
-}
-
-function walkInputs(
-	config: Config,
-	visitedTypes: Set<string>,
-	inputObj: InputObject,
-	rootType: graphql.TypeNode | graphql.GraphQLNamedType
-) {
-	// find the core type
-	const { type } = unwrapType(config, rootType)
-
-	// if we've seen this type already
-	if (visitedTypes.has(type.name)) {
-		// don't do anything else
-		return
-	}
-
-	// if this is a scalar or enum then we don't need to add anything to the type object
-	if (graphql.isEnumType(type) || graphql.isScalarType(type)) {
-		return
-	}
-	if (graphql.isUnionType(type)) {
-		return
-	}
-
-	// we haven't seen this type before and are about to generate the type
-	visitedTypes.add(type.name)
-
-	// generate the entry for the type
-	inputObj!.types[type.name] = Object.values(type.getFields()).reduce(
-		(typeFields, field: graphql.GraphQLInputField) => {
-			const { type: fieldType } = unwrapType(config, field.type)
-
-			// keep walking down
-			walkInputs(config, visitedTypes, inputObj, fieldType)
-
-			return {
-				...typeFields,
-				[field.name]: fieldType.toString(),
-			}
-		},
-		{}
-	)
-}