Parser (#9)

Reviewed-on: #9
Co-authored-by: phoenix <kundeng00@pm.me>
Co-committed-by: phoenix <kundeng00@pm.me>

.gitea/workflows/workflow.yaml

@@ -8,7 +8,7 @@ on:
 
 jobs:
   build:
-    runs-on: ubuntu-latest  # You can change this to macos-latest or windows-latest if needed
+    runs-on: ubuntu-24.04  # You can change this to macos-latest or windows-latest if needed
 
     steps:
     - uses: actions/checkout@v3
@@ -39,4 +39,3 @@ jobs:
           echo "Go code has style issues. Please run 'golint./...' to see them."
           exit 1
         fi
-

.gitignore

@@ -1 +1,2 @@
 /clean_file
+numbers.json

main.go

@@ -2,8 +2,9 @@ package main
 
 import "fmt"
 import "os"
+import "git.kundeng.us/phoenix/clean_file/parser"
 
-const VERSION = "0.4.101"
+const VERSION = "0.3.101"
 
 func main() {
 	fmt.Println("clean_file")
@@ -17,6 +18,13 @@ func main() {
 	filepath := args[1]
 
 	fmt.Println("File path:", filepath)
-	// TODO: Create parser package and Parser struct to do the remaining
+	prsr := parser.NumberParser{FilePath: filepath}
-	// work
+
+	rawValues := prsr.FileDump()
+	parsedValues := prsr.UpdateValues(rawValues)
+	newUpdated := prsr.RemoveDups(parsedValues)
+
+	prsr.PrintValues(newUpdated, true)
+
+	prsr.SaveFile(newUpdated)
 }

parser/parser.go

@@ -0,0 +1,173 @@
+package parser
+
+import "fmt"
+import "os"
+import "unicode"
+import (
+	"bufio"
+	"log"
+)
+import "encoding/json"
+import "strings"
+
+const Letters = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"
+
+type SomeObject struct {
+	Value string `json:"value"`
+}
+
+type NumberParser struct {
+	FilePath string
+}
+
+func (prsr NumberParser) FileDump() []SomeObject {
+	fmt.Println("Dumping files")
+	objs := []SomeObject{}
+
+	for _, line := range prsr.readLines(prsr.FilePath) {
+		if strings.ContainsAny(line, Letters) {
+			continue
+		}
+
+		myObj := SomeObject{Value: line}
+		if myObj.Value != "" {
+			objs = append(objs, myObj)
+		}
+	}
+
+	return objs
+}
+
+func (prsr NumberParser) UpdateValues(vals []SomeObject) []SomeObject {
+	updated := []SomeObject{}
+
+	for _, val := range vals {
+		parsed := val.Value
+		if len(parsed) < 2 {
+			fmt.Println("Invalid number:", parsed)
+			continue
+		}
+
+		pars := prsr.removeSomeData(parsed)
+		parsed = pars
+
+		fmt.Println(parsed)
+
+		updatedParsed := prsr.addPrefix(parsed)
+		updated = append(updated, SomeObject{Value: updatedParsed})
+	}
+
+	return updated
+}
+
+func (prsr NumberParser) RemoveDups(vals []SomeObject) []SomeObject {
+	updated := []SomeObject{}
+	uniqueItems := make(map[string]string)
+
+	for _, val := range vals {
+		uniqueItems[val.Value] = val.Value
+	}
+
+	for _, item := range uniqueItems {
+		updated = append(updated, SomeObject{Value: item})
+	}
+
+	return updated
+}
+
+func (prsr NumberParser) PrintValues(vals []SomeObject, printTotal bool) {
+	fmt.Println("Printing values")
+
+	for _, val := range vals {
+		fmt.Println("Value:", val.Value)
+	}
+
+	if printTotal {
+		fmt.Println("Total Numbers:", len(vals))
+	}
+}
+
+func (prsr NumberParser) SaveFile(vals []SomeObject) {
+	filename := "numbers.json"
+	allNumbers := []string{}
+
+	for _, val := range vals {
+		allNumbers = append(allNumbers, val.Value)
+	}
+
+	// Create json and save it to the filesystem
+	jsonData, err := json.MarshalIndent(allNumbers, "", "  ")
+	if err != nil {
+		fmt.Println("Error marshaling JSON:", err)
+		return
+	}
+
+	err = os.WriteFile(filename, jsonData, 0644)
+	if err != nil {
+		fmt.Println("Error writing file:", err)
+		return
+	}
+}
+
+func (prsr NumberParser) readLines(filename string) []string {
+	// Open the file
+	file, err := os.Open(filename)
+	if err != nil {
+		log.Fatal(err)
+	}
+	defer file.Close()
+
+	// Create a scanner to read the file line by line
+	scanner := bufio.NewScanner(file)
+	lines := []string{}
+
+	// Read line by line
+	for scanner.Scan() {
+		line := scanner.Text()
+		lines = append(lines, line)
+	}
+
+	// Check for any errors during scanning
+	if err := scanner.Err(); err != nil {
+		log.Fatal(err)
+	}
+
+	return lines
+}
+
+func (prsr NumberParser) removeSomeData(unparsed string) string {
+	parsed := ""
+
+	for _, ch := range unparsed {
+		if ch == ' ' {
+			continue
+		} else {
+			if unicode.IsDigit(ch) {
+				parsed += string(ch)
+			}
+		}
+	}
+
+	fmt.Println("Parsed:", parsed)
+
+	return parsed
+}
+
+func (prsr NumberParser) addPrefix(raw string) string {
+	parsed := raw
+
+	firstChar := raw[0]
+	secondChar := raw[1]
+
+	if firstChar == '1' {
+		parsed = "+" + parsed
+	} else if firstChar != '+' {
+		if secondChar != '1' {
+			parsed = "+1" + parsed
+		} else {
+			parsed = "+1" + parsed
+		}
+	}
+
+	return parsed
+}

parser/parser_test.go

@@ -0,0 +1,129 @@
+package parser
+
+import "testing"
+import "fmt"
+
+const Test_File_Path = "../tests/numbers.txt"
+
+// Add three tests
+// File dump
+func Test_FileDump(t *testing.T) {
+	parser := NumberParser{FilePath: Test_File_Path}
+	rawValues := parser.FileDump()
+
+	testValues := []SomeObject{
+		{Value: "(217) 555-0194"},
+		{Value: "(602) 555-0177"},
+		{Value: "(713) 555-0167"},
+		{Value: "(608) 555-0190"},
+		{Value: "(305) 555-0112"},
+		{Value: "(415) 555-0135"},
+		{Value: "(608) 555-0190"},
+		{Value: "(602) 555-0177"},
+		{Value: "(317) 555-0123"},
+		{Value: "(360) 555-0158"},
+	}
+
+	if len(rawValues) < 1 {
+		t.Error("Dump should not be empty")
+	}
+
+	if !dataEqual(rawValues, testValues) {
+		t.Error("File dump was not correctly dump")
+	}
+}
+
+// Parse values
+func Test_Parsed(t *testing.T) {
+	parser := NumberParser{FilePath: Test_File_Path}
+	rawValues := parser.FileDump()
+
+	if len(rawValues) < 1 {
+		t.Error("Dump should not be empty")
+	}
+
+	parsed := parser.UpdateValues(rawValues)
+	testValues := []SomeObject{
+		{Value: "+12175550194"},
+		{Value: "+16025550177"},
+		{Value: "+17135550167"},
+		{Value: "+16085550190"},
+		{Value: "+13055550112"},
+		{Value: "+14155550135"},
+		{Value: "+16085550190"},
+		{Value: "+16025550177"},
+		{Value: "+13175550123"},
+		{Value: "+13605550158"},
+	}
+
+	aSize := len(parsed)
+	bSize := len(testValues)
+	if aSize != bSize {
+		t.Error("Data has varrying sizes A", aSize, " B", bSize)
+	}
+	if !dataEqual(parsed, testValues) {
+		t.Error("Data has not be updated")
+	}
+}
+
+// Remove duplicates
+func Test_RemoveDups(t *testing.T) {
+	parser := NumberParser{FilePath: Test_File_Path}
+	rawValues := parser.FileDump()
+
+	if len(rawValues) < 1 {
+		t.Error("Dump should not be empty")
+	}
+
+	parsed := parser.UpdateValues(rawValues)
+	finalAmount := parser.RemoveDups(parsed)
+	testValues := []SomeObject{
+		{Value: "+12175550194"},
+		{Value: "+16025550177"},
+		{Value: "+17135550167"},
+		{Value: "+16085550190"},
+		{Value: "+13055550112"},
+		{Value: "+14155550135"},
+		{Value: "+13175550123"},
+		{Value: "+13605550158"},
+	}
+
+	aSize := len(finalAmount)
+	bSize := len(testValues)
+	if aSize != bSize {
+		t.Error("Data has varrying sizes A", aSize, " B", bSize)
+	}
+	if !dataEqual(finalAmount, testValues) {
+		fmt.Println("Printing final amount")
+		for _, val := range finalAmount {
+			fmt.Println("Value:", val.Value)
+		}
+		fmt.Println("Printing test values")
+		for _, val := range testValues {
+			fmt.Println("Value:", val.Value)
+		}
+
+		t.Error("Data has not be updated")
+	}
+}
+
+func dataEqual(a, b []SomeObject) bool {
+	if len(a) != len(b) {
+		return false
+	}
+
+	for i := range a {
+		foundIt := false
+		for j := range b {
+			if a[i] == b[j] {
+				foundIt = true
+			}
+		}
+
+		if !foundIt {
+			return false
+		}
+	}
+
+	return true
+}

tests/numbers.txt

@@ -0,0 +1,10 @@
+(217) 555-0194
+(602) 555-0177
+(713) 555-0167
+(608) 555-0190
+(305) 555-0112
+(415) 555-0135
+(608) 555-0190
+(602) 555-0177
+(317) 555-0123
+(360) 555-0158