From 764b6a3a2c89d0e4d948862bfeb546d33bbf45e6 Mon Sep 17 00:00:00 2001 From: "(quasar) nebula" Date: Tue, 14 Nov 2023 21:45:35 -0400 Subject: data: generic composite dependency comments --- src/data/composite/wiki-data/index.js | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'src/data/composite/wiki-data') diff --git a/src/data/composite/wiki-data/index.js b/src/data/composite/wiki-data/index.js index 1d0400fc..df50a2db 100644 --- a/src/data/composite/wiki-data/index.js +++ b/src/data/composite/wiki-data/index.js @@ -1,6 +1,13 @@ +// #composite/wiki-data +// +// Entries here may depend on entries in #composite/control-flow and in +// #composite/data. +// + export {default as exitWithoutContribs} from './exitWithoutContribs.js'; export {default as inputThingClass} from './inputThingClass.js'; export {default as inputWikiData} from './inputWikiData.js'; +export {default as withParsedCommentaryEntries} from './withParsedCommentaryEntries.js'; export {default as withResolvedContribs} from './withResolvedContribs.js'; export {default as withResolvedReference} from './withResolvedReference.js'; export {default as withResolvedReferenceList} from './withResolvedReferenceList.js'; -- cgit 1.3.0-6-gf8a5 From 6deea0629a3f3b9985d205d2f3a048893ea938c9 Mon Sep 17 00:00:00 2001 From: "(quasar) nebula" Date: Tue, 14 Nov 2023 22:20:45 -0400 Subject: data, test: withParsedCommentaryEntries --- .../wiki-data/withParsedCommentaryEntries.js | 181 +++++++++++++++++++++ 1 file changed, 181 insertions(+) create mode 100644 src/data/composite/wiki-data/withParsedCommentaryEntries.js (limited to 'src/data/composite/wiki-data') diff --git a/src/data/composite/wiki-data/withParsedCommentaryEntries.js b/src/data/composite/wiki-data/withParsedCommentaryEntries.js new file mode 100644 index 00000000..5bd72dc9 --- /dev/null +++ b/src/data/composite/wiki-data/withParsedCommentaryEntries.js @@ -0,0 +1,181 @@ +import {input, templateCompositeFrom} from '#composite'; +import find from '#find'; +import {stitchArrays} from '#sugar'; +import {isCommentary} from '#validators'; + +import {fillMissingListItems, withPropertiesFromList} from '#composite/data'; + +import withResolvedReferenceList from './withResolvedReferenceList.js'; + +// Matches in roughly the format: +// +// artistReference: (annotation, date) +// +// where capturing group "annotation" can be any text at all, except that the +// last entry (past a comma or the only content within parentheses), if parsed +// as a date, is the capturing group "date". "Parsing as a date" means one of +// these formats: +// +// * "25 December 2019" - one or two number digits, followed by any text, +// followed by four number digits +// * "12/25/2019" - one or two number digits, a slash, one or two number +// digits, a slash, and two to four number digits +// +// The artist reference can optionally be boldface (in ), which will be +// captured as non-null in "boldfaceArtist". Otherwise it is all the characters +// between and and is captured in "artistReference" and is either the +// name of an artist or an "artist:directory"-style reference. +// +export const commentaryRegex = + /^(?)?(?.+):(?:<\/b>)?<\/i>(?: \((?(?:.*?(?=[,)]))*?)(?:,? ?(?[0-9]{1,2} [^,]*[0-9]{4,4}|[0-9]{1,2}\/[0-9]{1,2}\/[0-9]{2,4}))?\))?/gm; + +export default templateCompositeFrom({ + annotation: `withParsedCommentaryEntries`, + + inputs: { + from: input({validate: isCommentary}), + }, + + outputs: ['#parsedCommentaryEntries'], + + steps: () => [ + { + dependencies: [input('from')], + + compute: (continuation, { + [input('from')]: commentaryText, + }) => continuation({ + ['#rawMatches']: + Array.from(commentaryText.matchAll(commentaryRegex)), + }), + }, + + withPropertiesFromList({ + list: '#rawMatches', + properties: input.value([ + '0', // The entire match as a string. + 'groups', + 'index', + ]), + }).outputs({ + '#rawMatches.0': '#rawMatches.text', + '#rawMatches.groups': '#rawMatches.groups', + '#rawMatches.index': '#rawMatches.startIndex', + }), + + { + dependencies: [ + '#rawMatches.text', + '#rawMatches.startIndex', + ], + + compute: (continuation, { + ['#rawMatches.text']: text, + ['#rawMatches.startIndex']: startIndex, + }) => continuation({ + ['#rawMatches.endIndex']: + stitchArrays({text, startIndex}) + .map(({text, startIndex}) => startIndex + text.length), + }), + }, + + { + dependencies: [ + input('from'), + '#rawMatches.startIndex', + '#rawMatches.endIndex', + ], + + compute: (continuation, { + [input('from')]: commentaryText, + ['#rawMatches.startIndex']: startIndex, + ['#rawMatches.endIndex']: endIndex, + }) => continuation({ + ['#entries.body']: + stitchArrays({startIndex, endIndex}) + .map(({endIndex}, index, stitched) => + (index === stitched.length - 1 + ? commentaryText.slice(endIndex) + : commentaryText.slice( + endIndex, + stitched[index + 1].startIndex))) + .map(body => body.trim()), + }), + }, + + withPropertiesFromList({ + list: '#rawMatches.groups', + prefix: input.value('#entries'), + properties: input.value([ + 'artistReference', + 'boldfaceArtist', + 'annotation', + 'date', + ]), + }), + + // The artistReference group will always have a value, since it's required + // for the line to match in the first place. + + withResolvedReferenceList({ + list: '#entries.artistReference', + data: 'artistData', + find: input.value(find.artist), + notFoundMode: input.value('null'), + }).outputs({ + '#resolvedReferenceList': '#entries.artist', + }), + + { + dependencies: ['#entries.boldfaceArtist'], + compute: (continuation, { + ['#entries.boldfaceArtist']: boldfaceArtist, + }) => continuation({ + ['#entries.boldfaceArtist']: + boldfaceArtist.map(boldface => boldface ? true : false), + }), + }, + + fillMissingListItems({ + list: '#entries.annotation', + fill: input.value(null), + }), + + { + dependencies: ['#entries.date'], + compute: (continuation, { + ['#entries.date']: date, + }) => continuation({ + ['#entries.date']: + date.map(date => date ? new Date(date) : null), + }), + }, + + { + dependencies: [ + '#entries.artist', + '#entries.boldfaceArtist', + '#entries.annotation', + '#entries.date', + '#entries.body', + ], + + compute: (continuation, { + ['#entries.artist']: artist, + ['#entries.boldfaceArtist']: boldfaceArtist, + ['#entries.annotation']: annotation, + ['#entries.date']: date, + ['#entries.body']: body, + }) => continuation({ + ['#parsedCommentaryEntries']: + stitchArrays({ + artist, + boldfaceArtist, + annotation, + date, + body, + }), + }), + }, + ], +}); -- cgit 1.3.0-6-gf8a5 From 362dc0619b93d74ad34df1bfbfd9ebc632fa5156 Mon Sep 17 00:00:00 2001 From: "(quasar) nebula" Date: Tue, 14 Nov 2023 22:49:51 -0400 Subject: data, yaml: catch commentary artist ref errors --- .../wiki-data/withParsedCommentaryEntries.js | 23 +--------------------- 1 file changed, 1 insertion(+), 22 deletions(-) (limited to 'src/data/composite/wiki-data') diff --git a/src/data/composite/wiki-data/withParsedCommentaryEntries.js b/src/data/composite/wiki-data/withParsedCommentaryEntries.js index 5bd72dc9..9e33cdac 100644 --- a/src/data/composite/wiki-data/withParsedCommentaryEntries.js +++ b/src/data/composite/wiki-data/withParsedCommentaryEntries.js @@ -2,33 +2,12 @@ import {input, templateCompositeFrom} from '#composite'; import find from '#find'; import {stitchArrays} from '#sugar'; import {isCommentary} from '#validators'; +import {commentaryRegex} from '#wiki-data'; import {fillMissingListItems, withPropertiesFromList} from '#composite/data'; import withResolvedReferenceList from './withResolvedReferenceList.js'; -// Matches in roughly the format: -// -// artistReference: (annotation, date) -// -// where capturing group "annotation" can be any text at all, except that the -// last entry (past a comma or the only content within parentheses), if parsed -// as a date, is the capturing group "date". "Parsing as a date" means one of -// these formats: -// -// * "25 December 2019" - one or two number digits, followed by any text, -// followed by four number digits -// * "12/25/2019" - one or two number digits, a slash, one or two number -// digits, a slash, and two to four number digits -// -// The artist reference can optionally be boldface (in ), which will be -// captured as non-null in "boldfaceArtist". Otherwise it is all the characters -// between and and is captured in "artistReference" and is either the -// name of an artist or an "artist:directory"-style reference. -// -export const commentaryRegex = - /^(?)?(?.+):(?:<\/b>)?<\/i>(?: \((?(?:.*?(?=[,)]))*?)(?:,? ?(?[0-9]{1,2} [^,]*[0-9]{4,4}|[0-9]{1,2}\/[0-9]{1,2}\/[0-9]{2,4}))?\))?/gm; - export default templateCompositeFrom({ annotation: `withParsedCommentaryEntries`, -- cgit 1.3.0-6-gf8a5 From f754a8d9187e435a761db31b5053aa2e7ba22e13 Mon Sep 17 00:00:00 2001 From: "(quasar) nebula" Date: Tue, 14 Nov 2023 23:36:37 -0400 Subject: data, test: boldfaceArtist -> artistDisplayText --- .../wiki-data/withParsedCommentaryEntries.js | 21 ++++++++------------- 1 file changed, 8 insertions(+), 13 deletions(-) (limited to 'src/data/composite/wiki-data') diff --git a/src/data/composite/wiki-data/withParsedCommentaryEntries.js b/src/data/composite/wiki-data/withParsedCommentaryEntries.js index 9e33cdac..7b1c9484 100644 --- a/src/data/composite/wiki-data/withParsedCommentaryEntries.js +++ b/src/data/composite/wiki-data/withParsedCommentaryEntries.js @@ -87,7 +87,7 @@ export default templateCompositeFrom({ prefix: input.value('#entries'), properties: input.value([ 'artistReference', - 'boldfaceArtist', + 'artistDisplayText', 'annotation', 'date', ]), @@ -105,15 +105,10 @@ export default templateCompositeFrom({ '#resolvedReferenceList': '#entries.artist', }), - { - dependencies: ['#entries.boldfaceArtist'], - compute: (continuation, { - ['#entries.boldfaceArtist']: boldfaceArtist, - }) => continuation({ - ['#entries.boldfaceArtist']: - boldfaceArtist.map(boldface => boldface ? true : false), - }), - }, + fillMissingListItems({ + list: '#entries.artistDisplayText', + fill: input.value(null), + }), fillMissingListItems({ list: '#entries.annotation', @@ -133,7 +128,7 @@ export default templateCompositeFrom({ { dependencies: [ '#entries.artist', - '#entries.boldfaceArtist', + '#entries.artistDisplayText', '#entries.annotation', '#entries.date', '#entries.body', @@ -141,7 +136,7 @@ export default templateCompositeFrom({ compute: (continuation, { ['#entries.artist']: artist, - ['#entries.boldfaceArtist']: boldfaceArtist, + ['#entries.artistDisplayText']: artistDisplayText, ['#entries.annotation']: annotation, ['#entries.date']: date, ['#entries.body']: body, @@ -149,7 +144,7 @@ export default templateCompositeFrom({ ['#parsedCommentaryEntries']: stitchArrays({ artist, - boldfaceArtist, + artistDisplayText, annotation, date, body, -- cgit 1.3.0-6-gf8a5 From a34b8d027866fbe858a4d2ff3543bc84c9d5983a Mon Sep 17 00:00:00 2001 From: "(quasar) nebula" Date: Fri, 17 Nov 2023 06:53:34 -0400 Subject: data, yaml, content: support multiple artists per commentary entry --- .../wiki-data/withParsedCommentaryEntries.js | 41 +++++++++++++++++----- 1 file changed, 33 insertions(+), 8 deletions(-) (limited to 'src/data/composite/wiki-data') diff --git a/src/data/composite/wiki-data/withParsedCommentaryEntries.js b/src/data/composite/wiki-data/withParsedCommentaryEntries.js index 7b1c9484..25c07a37 100644 --- a/src/data/composite/wiki-data/withParsedCommentaryEntries.js +++ b/src/data/composite/wiki-data/withParsedCommentaryEntries.js @@ -4,7 +4,12 @@ import {stitchArrays} from '#sugar'; import {isCommentary} from '#validators'; import {commentaryRegex} from '#wiki-data'; -import {fillMissingListItems, withPropertiesFromList} from '#composite/data'; +import { + fillMissingListItems, + withFlattenedList, + withPropertiesFromList, + withUnflattenedList, +} from '#composite/data'; import withResolvedReferenceList from './withResolvedReferenceList.js'; @@ -86,23 +91,43 @@ export default templateCompositeFrom({ list: '#rawMatches.groups', prefix: input.value('#entries'), properties: input.value([ - 'artistReference', + 'artistReferences', 'artistDisplayText', 'annotation', 'date', ]), }), - // The artistReference group will always have a value, since it's required + // The artistReferences group will always have a value, since it's required // for the line to match in the first place. + { + dependencies: ['#entries.artistReferences'], + compute: (continuation, { + ['#entries.artistReferences']: artistReferenceTexts, + }) => continuation({ + ['#entries.artistReferences']: + artistReferenceTexts + .map(text => text.split(',').map(ref => ref.trim())), + }), + }, + + withFlattenedList({ + list: '#entries.artistReferences', + }), + withResolvedReferenceList({ - list: '#entries.artistReference', + list: '#flattenedList', data: 'artistData', find: input.value(find.artist), notFoundMode: input.value('null'), + }), + + withUnflattenedList({ + list: '#resolvedReferenceList', + filter: input.value(false), }).outputs({ - '#resolvedReferenceList': '#entries.artist', + '#unflattenedList': '#entries.artists', }), fillMissingListItems({ @@ -127,7 +152,7 @@ export default templateCompositeFrom({ { dependencies: [ - '#entries.artist', + '#entries.artists', '#entries.artistDisplayText', '#entries.annotation', '#entries.date', @@ -135,7 +160,7 @@ export default templateCompositeFrom({ ], compute: (continuation, { - ['#entries.artist']: artist, + ['#entries.artists']: artists, ['#entries.artistDisplayText']: artistDisplayText, ['#entries.annotation']: annotation, ['#entries.date']: date, @@ -143,7 +168,7 @@ export default templateCompositeFrom({ }) => continuation({ ['#parsedCommentaryEntries']: stitchArrays({ - artist, + artists, artistDisplayText, annotation, date, -- cgit 1.3.0-6-gf8a5 From 5b8060bb86d457a0d23b607aa866c4d7d6eb6f0f Mon Sep 17 00:00:00 2001 From: "(quasar) nebula" Date: Sat, 18 Nov 2023 20:16:40 -0400 Subject: data: withParsedCommentaryEntries: filter out null artists --- src/data/composite/wiki-data/withParsedCommentaryEntries.js | 1 - 1 file changed, 1 deletion(-) (limited to 'src/data/composite/wiki-data') diff --git a/src/data/composite/wiki-data/withParsedCommentaryEntries.js b/src/data/composite/wiki-data/withParsedCommentaryEntries.js index 25c07a37..edfc9e3c 100644 --- a/src/data/composite/wiki-data/withParsedCommentaryEntries.js +++ b/src/data/composite/wiki-data/withParsedCommentaryEntries.js @@ -125,7 +125,6 @@ export default templateCompositeFrom({ withUnflattenedList({ list: '#resolvedReferenceList', - filter: input.value(false), }).outputs({ '#unflattenedList': '#entries.artists', }), -- cgit 1.3.0-6-gf8a5 From 5c1d9fb97b8ecc61c0343d6eee63a735c34e53c9 Mon Sep 17 00:00:00 2001 From: "(quasar) nebula" Date: Sun, 3 Dec 2023 13:25:05 -0400 Subject: data: withThingsSortedAlphabetically --- src/data/composite/wiki-data/index.js | 1 + .../wiki-data/withThingsSortedAlphabetically.js | 122 +++++++++++++++++++++ 2 files changed, 123 insertions(+) create mode 100644 src/data/composite/wiki-data/withThingsSortedAlphabetically.js (limited to 'src/data/composite/wiki-data') diff --git a/src/data/composite/wiki-data/index.js b/src/data/composite/wiki-data/index.js index df50a2db..a2ff09d8 100644 --- a/src/data/composite/wiki-data/index.js +++ b/src/data/composite/wiki-data/index.js @@ -12,3 +12,4 @@ export {default as withResolvedContribs} from './withResolvedContribs.js'; export {default as withResolvedReference} from './withResolvedReference.js'; export {default as withResolvedReferenceList} from './withResolvedReferenceList.js'; export {default as withReverseReferenceList} from './withReverseReferenceList.js'; +export {default as withThingsSortedAlphabetically} from './withThingsSortedAlphabetically.js'; diff --git a/src/data/composite/wiki-data/withThingsSortedAlphabetically.js b/src/data/composite/wiki-data/withThingsSortedAlphabetically.js new file mode 100644 index 00000000..d2487e42 --- /dev/null +++ b/src/data/composite/wiki-data/withThingsSortedAlphabetically.js @@ -0,0 +1,122 @@ +// Sorts a list of live, generic wiki data objects alphabetically. +// Note that this uses localeCompare but isn't specialized to a particular +// language; where localization is concerned (in content), a follow-up, locale- +// specific sort should be performed. But this function does serve to organize +// a list so same-name entries are beside each other. + +import {input, templateCompositeFrom} from '#composite'; +import {validateWikiData} from '#validators'; +import {compareCaseLessSensitive, normalizeName} from '#wiki-data'; + +import {raiseOutputWithoutDependency} from '#composite/control-flow'; +import {withMappedList, withSortedList, withPropertiesFromList} + from '#composite/data'; + +export default templateCompositeFrom({ + annotation: `withThingsSortedAlphabetically`, + + inputs: { + things: input({validate: validateWikiData}), + }, + + outputs: ['#sortedThings'], + + steps: () => [ + raiseOutputWithoutDependency({ + dependency: input('things'), + mode: input.value('empty'), + output: input.value({'#sortedThings': []}), + }), + + withPropertiesFromList({ + list: input('things'), + properties: input.value(['name', 'directory']), + }).outputs({ + '#list.name': '#names', + '#list.directory': '#directories', + }), + + withMappedList({ + list: '#names', + map: input.value(normalizeName), + }).outputs({ + '#mappedList': '#normalizedNames', + }), + + withSortedList({ + list: '#normalizedNames', + sort: input.value(compareCaseLessSensitive), + }).outputs({ + '#unstableSortIndices': '#normalizedNameSortIndices', + }), + + withSortedList({ + list: '#names', + sort: input.value(compareCaseLessSensitive), + }).outputs({ + '#unstableSortIndices': '#nonNormalizedNameSortIndices', + }), + + withSortedList({ + list: '#directories', + sort: input.value(compareCaseLessSensitive), + }).outputs({ + '#unstableSortIndices': '#directorySortIndices', + }), + + // TODO: No primitive for the next two-three steps, yet... + + { + dependencies: [input('things')], + compute: (continuation, { + [input('things')]: things, + }) => continuation({ + ['#combinedSortIndices']: + Array.from( + {length: things.length}, + (_item, index) => index), + }), + }, + + { + dependencies: [ + '#combinedSortIndices', + '#normalizedNameSortIndices', + '#nonNormalizedNameSortIndices', + '#directorySortIndices', + ], + + compute: (continuation, { + ['#combinedSortIndices']: combined, + ['#normalizedNameSortIndices']: normalized, + ['#nonNormalizedNameSortIndices']: nonNormalized, + ['#directorySortIndices']: directory, + }) => continuation({ + ['#combinedSortIndices']: + combined.sort((index1, index2) => { + if (normalized[index1] !== normalized[index2]) + return normalized[index1] - normalized[index2]; + + if (nonNormalized[index1] !== nonNormalized[index2]) + return nonNormalized[index1] - nonNormalized[index2]; + + if (directory[index1] !== directory[index2]) + return directory[index1] - directory[index2]; + + return 0; + }), + }), + }, + + { + dependencies: [input('things'), '#combinedSortIndices'], + compute: (continuation, { + [input('things')]: things, + ['#combinedSortIndices']: combined, + }) => continuation({ + ['#sortedThings']: + combined.map(index => things[index]), + }), + }, + ], +}); -- cgit 1.3.0-6-gf8a5