feat(2023-01): don't use string replacement to parse text

developher-net · developher-net · commit 28efcb7f0926 · 2023-12-04T22:22:19.000-08:00
"eightwo" should render as "82" not "8wo"

Solves part 2
diff --git a/2023/day-01/checksum.js b/2023/day-01/checksum.js
@@ -14,34 +14,104 @@ const checksumLine = (data) => {
   return parseInt(result)
 }
 
+const lazyNums = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', 'nine']
+const lazyReg = new RegExp(lazyNums.join('|'))
+const lazyNumsReversed = lazyNums.map((num) => num.split('').reverse().join(''))
+const lazyReversReg = new RegExp(lazyNumsReversed.join('|'))
+
+const lookupPosition = (match) => {
+  const idx = lazyNums.indexOf(match)
+  if (idx > 9) {
+    return idx - 9
+  }
+  return idx
+}
+
+const lookupPositionReversed = (match) => {
+  const reverseMatch = match.split('').reverse().join('')
+  const idx = lazyNums.indexOf(reverseMatch)
+  if (idx > 9) {
+    return idx - 9
+  }
+  return idx
+}
+
+const lazyChecksumLine = (data) => {
+  let first = ''
+  data.replace(lazyReg, (match) => {
+    first = lookupPosition(match)
+    return match // reinsert so we don't bork the data string
+  })
+  // find last matching digit by reversing the string and searching backwards
+  let last = ''
+  data = data.split('').reverse().join('')
+  data.replace(lazyReversReg, (match) => {
+    last = lookupPositionReversed(match)
+    return match // reinsert so we don't bork the data string
+  })
+
+  return parseInt(`${first}${last}`)
+}
+
 /**
  * Generates the checksum for an entire set
  * @param {array} set of lines containing data
  */
-const checksumSet = (set) => {
+const checksumSet = (set, sanitize = false, lazy = false) => {
+  let filter = (data) => data
+  if (sanitize) {
+    filter = sanitizeLine
+  }
+
+  let checksum = checksumLine
+  if (lazy) {
+    checksum = lazyChecksumLine
+  }
+
   return set.reduce((total, current) => {
-    return total + checksumLine((current))
+    return total + checksum(
+      filter(current)
+    )
   }, 0)
 }
 
-/**
- * Generates the checksum for an entire set when data is not sanitized
- * @param {array} set of lines containing data
- */
-const checksumUnSanitizedSet = (set) => {
-  return set.reduce((total, current) => {
-    return total + checksumLine(sanitizeLine(current))
-  }, 0)
+const numbers = ['one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', 'nine']
+const numbersReversed = numbers.map((num) => num.split('').reverse().join(''))
+const reg = new RegExp(numbers.join('|'))
+const regGlobal = new RegExp(numbers.join('|'), 'g')
+const regReversed = new RegExp(numbersReversed.join('|'))
+
+// Sanitizes using a single-pass regex replace all
+// Produces 53885 which is incorrect for part 2
+const byRegex = (data) => {
+  return data.replaceAll(regGlobal, (matched) => numbers.indexOf(matched) + 1)
+}
+
+// Sanitizes by replacing just the first and last text digits, in case shared letters is supposed to work
+// Produces 53853 which is too low for part 2
+const byFirstLast = (data) => {
+  // sanitize first matching digit
+  data = data.replace(reg, (matched) => numbers.indexOf(matched) + 1)
+  // sanitize last matching digit by reversing the string and searching backwards
+  data = data.split('').reverse().join('')
+  data = data.replace(regReversed, (matched) => numbersReversed.indexOf(matched) + 1)
+
+  // return original order
+  return data.split('').reverse().join('')
 }
 
-const numbers = ['zero', 'one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', 'nine']
-const reg = new RegExp(numbers.join('|'), 'g')
 /**
  * Sanitzizes a line by replacing spelled-out numbers with data
  * @param {string} data line of input to sanitize
  */
-const sanitizeLine = (data) => {
-  return data.replaceAll(reg, (matched) => numbers.indexOf(matched))
+const sanitizeLine = (data, method = 'byFirstLast') => {
+  const methods = {
+    none: (data) => data,
+    byFirstLast,
+    byRegex
+  }
+
+  return methods[method](data)
 }
 
-module.exports = { checksumLine, checksumSet, checksumUnSanitizedSet, sanitizeLine }
+module.exports = { checksumLine, checksumSet, sanitizeLine, lazyChecksumLine }
diff --git a/2023/day-01/checksum.test.js b/2023/day-01/checksum.test.js
@@ -1,6 +1,6 @@
 /* eslint-env mocha */
 const { expect } = require('chai')
-const { checksumSet, checksumUnSanitizedSet, checksumLine, sanitizeLine } = require('./checksum')
+const { checksumSet, checksumLine, sanitizeLine, lazyChecksumLine } = require('./checksum')
 const fs = require('fs')
 const path = require('path')
 const filePath = path.join(__dirname, 'input.txt')
@@ -30,7 +30,7 @@ describe('--- Day 1: Trebuchet?! ---', () => {
   })
   describe('Part 2', () => {
     describe('sanitizeLine', () => {
-      const set = [
+      const data = [
         'two1nine',
         'eightwothree',
         'abcone2threexyz',
@@ -41,40 +41,119 @@ describe('--- Day 1: Trebuchet?! ---', () => {
       ]
       const result = [29, 83, 13, 24, 42, 14, 76]
       it('cleans up a string when digits are spelled out', () => {
+        const set = JSON.parse(JSON.stringify(data))
         for (let x = 0; x < set.length; x++) {
           expect(checksumLine(sanitizeLine(set[x]))).to.equal(result[x])
+          // expect(checksumLine(sanitizeLine(set[x], 'sanitizeByRegex'))).to.equal(result[x])
+          // expect(checksumLine(sanitizeLine(set[x], 'sanitizeFirstLast'))).to.equal(result[x])
         }
       })
-      it('handles first matches, and doesn\'t allow for multiple words to share letters', () => {
-        expect(sanitizeLine('eightwothree')).to.equal('8wo3')
+      it('allows for skipping sanitation', () => {
+        const set = JSON.parse(JSON.stringify(data))
+        for (let x = 0; x < set.length; x++) {
+          expect(sanitizeLine(set[x], 'none')).to.equal(data[x])
+        }
       })
     })
-    describe('checksumUnSanitizedSet', () => {
-      it('calculates the checksum for a set of lines by summing the checksum of each sanitized line', () => {
-        const set = [
-          'two1nine',
-          'eightwothree',
-          'abcone2threexyz',
-          'xtwone3four',
-          '4nineeightseven2',
-          'zoneight234',
-          '7pqrstsixteen'
-        ]
-        expect(checksumUnSanitizedSet(set)).to.equal(281)
+    describe('checksumSet', () => {
+      const data = [
+        'two1nine',
+        'eightwothree',
+        'abcone2threexyz',
+        'xtwone3four',
+        '4nineeightseven2',
+        'zoneight234',
+        '7pqrstsixteen'
+      ]
+      it('can sanitize', () => {
+        expect(checksumSet(data, true)).to.equal(281)
       })
     })
+    describe('lazyChecksumLine', () => {
+      const data = [
+        'two1nine',
+        'eightwothree',
+        'abcone2threexyz',
+        'xtwone3four',
+        '4nineeightseven2',
+        'zoneight234',
+        '7pqrstsixteen'
+      ]
+      const result = [29, 83, 13, 24, 42, 14, 76]
+      it('can match text or numeric for checksum calcs', () => {
+        const set = JSON.parse(JSON.stringify(data))
+        for (let x = 0; x < set.length; x++) {
+          expect(lazyChecksumLine(set[x])).to.equal(result[x])
+        }
+      })
+    })
+
+    //   it('doesn`t sanitize by default', () => {
+    //     const set = [
+    //       'two1nine',
+    //       'eightwothree',
+    //       'abcone2threexyz',
+    //       'xtwone3four',
+    //       '4nineeightseven2',
+    //       'zoneight234',
+    //       '7pqrstsixteen'
+    //     ]
+    //     expect(checksumSet(set)).to.be.NaN
+    //   })
+    //   it('allows for sanitizing to be explicitly disabled', () => {
+    //     const set = [
+    //       'two1nine',
+    //       'eightwothree',
+    //       'abcone2threexyz',
+    //       'xtwone3four',
+    //       '4nineeightseven2',
+    //       'zoneight234',
+    //       '7pqrstsixteen'
+    //     ]
+    //     expect(checksumSet(set, 'none')).to.be.NaN
+    //   })
+    //   // it('calculates the checksum for a set of lines by summing the checksum of each line', () => {
+    //   //   const set = [
+    //   //     'two1nine',
+    //   //     'eightwothree',
+    //   //     'abcone2threexyz',
+    //   //     'xtwone3four',
+    //   //     '4nineeightseven2',
+    //   //     'zoneight234',
+    //   //     '7pqrstsixteen'
+    //   //   ]
+    //   //   expect(checksumSet(set)).to.equal(281)
+    //   // })
+    // })
     describe('integeration', () => {
-      it('53853 is too low for part 2', (done) => {
-        let data
-        fs.readFile(filePath, { encoding: 'utf8' }, (err, initData) => {
+      let initData
+      before((done) => {
+        fs.readFile(filePath, { encoding: 'utf8' }, (err, rawData) => {
           if (err) throw err
-          initData = inputToArray(initData.trim())
+          initData = inputToArray(rawData.trim())
           // Deep copy to ensure we aren't mutating the original data
-          data = JSON.parse(JSON.stringify(initData))
-          expect(checksumUnSanitizedSet(data)).to.be.greaterThan(53853)
+          // data = JSON.parse(JSON.stringify(initData))
           done()
         })
       })
+
+      it('is not done without sanitation, since that matches part 1', () => {
+        const data = JSON.parse(JSON.stringify(initData))
+        const result = checksumSet(data, true)
+        expect(result).to.not.equal(54953)
+      })
+
+      it('is not done with a one-time regex', () => {
+        const data = JSON.parse(JSON.stringify(initData))
+        const result = checksumSet(data, true)
+        expect(result).to.not.equal(53885) // result of one-time regex
+      })
+
+      it('is not done by sanitizing just the first and last strings', () => {
+        const data = JSON.parse(JSON.stringify(initData))
+        const result = checksumSet(data, true)
+        expect(result).to.not.equal(53853) // result of first/last substitution onlye
+      })
     })
   })
 })
diff --git a/2023/day-01/solution.js b/2023/day-01/solution.js
@@ -21,8 +21,7 @@ fs.readFile(filePath, { encoding: 'utf8' }, (err, initData) => {
 
   const part2 = () => {
     const data = resetInput()
-    console.debug(data)
-    return 'No answer yet'
+    return checksumSet(data, false, true)
   }
   const answers = []
   answers.push(part1())

Original file line number	Diff line number	Diff line change
`@@ -21,8 +21,7 @@ fs.readFile(filePath, { encoding: 'utf8' }, (err, initData) => {`
`21`	`21`
`22`	`22`	`const part2 = () => {`
`23`	`23`	`const data = resetInput()`
`24`		`- console.debug(data)`
`25`		`- return 'No answer yet'`
	`24`	`+ return checksumSet(data, false, true)`
`26`	`25`	`}`
`27`	`26`	`const answers = []`
`28`	`27`	`answers.push(part1())`