Parse a text file and split sections based on vertical anchors

vertSplitter.go

// vertSplitter parses an AWS credentials file and splits the sections
// up based on profiles. It then parses each section to build a
// JSON profile struct for each profile including comments.

package main

import (
	"bufio"
	"encoding/json"
	"fmt"
	"os"
	"regexp"
)

// readFileAnchorChunks takes in input filename and regex string
// it then reads the file and breaks the contents up in to a map of
// indexed string slices based on the anchor provided. It's essentially
// vertically splitting a file based on a regex separator.
func readFileAnchorChunks(filename string, anchorRegex string) (chunks map[int][]string, err error) {
	chunks = make(map[int][]string)
	f, err := os.OpenFile(filename, os.O_RDONLY, os.ModeAppend)
	if err != nil {
		return chunks, err
	}
	scanner := bufio.NewScanner(f)
	reAnchor, err := regexp.Compile(anchorRegex)
	if err != nil {
		return chunks, err
	}
	var data []string
	for scanner.Scan() {
		data = append(data, scanner.Text())
	}
	// first find the anchor index points
	var anchors []int
	for index, line := range data {
		matchEntry := reAnchor.FindStringSubmatch(line)
		if len(matchEntry) > 0 {
			anchors = append(anchors, index)
		}
	}
	// now walk through the data between anchor lines
	for anchor, linenum := range anchors {
		if (anchor + 1) > len(anchors)-1 {
			// means we're at the last anchor and have to range differently
			for i := linenum; i <= len(data)-1; i++ {
				chunks[anchor] = append(chunks[anchor], data[i])
			}
		} else {
			for i := linenum; i <= anchors[anchor+1]-1; i++ {
				chunks[anchor] = append(chunks[anchor], data[i])
			}
		}
	}
	return chunks, err
}

func subExSingleFinder(line string, reOb *regexp.Regexp, matchGroup string) (resultSingle string, found bool) {
	match := reOb.FindStringSubmatch(line)
	found = false
	if len(match) > 0 {
		for i, name := range reOb.SubexpNames() {
			if i != 0 {
				if name == matchGroup {
					resultSingle = match[i]
					found = true
					return resultSingle, found
				}
			}
		}
	}
	return resultSingle, found
}

func loadProfileFile(filename string) (profiles []Profile, err error) {
	reProfileRString := `^\[(?P<profilename>.*)\]`
	chunks, err := readFileAnchorChunks(filename, reProfileRString)
	if err != nil {
		return profiles, err
	}
	for _, chunk := range chunks {
		// now that we have a pre-separated chunk we can assume it's
		// a new profile entry
		var p Profile
		for _, line := range chunk {
			reProfile := regexp.MustCompile(reProfileRString)
			reComment := regexp.MustCompile("^#(?P<comment>.*)")
			reOutput := regexp.MustCompile("^output = (?P<output>.*)")
			reRegion := regexp.MustCompile("^region = (?P<region>.*)")
			reAKID := regexp.MustCompile("^aws_access_key_id = (?P<akid>.*)")
			reSAK := regexp.MustCompile("^aws_secret_access_key = (?P<sak>.*)")
			reST := regexp.MustCompile("^aws_session_token = (?P<st>.*)")
			reRole := regexp.MustCompile(".*ASSUMED ROLE: (?P<rolearn>.*)")
			matchComment := reComment.FindStringSubmatch(line)
			if len(matchComment) > 0 {
				p.Comments = append(p.Comments, matchComment[0])
			}
			result, found := subExSingleFinder(line, reProfile, "profilename")
			if found {
				p.ProfileName = result
			}
			result, found = subExSingleFinder(line, reOutput, "output")
			if found {
				p.Output = result
			}
			result, found = subExSingleFinder(line, reRegion, "region")
			if found {
				p.Region = result
			}
			result, found = subExSingleFinder(line, reAKID, "akid")
			if found {
				p.AWSAccessKeyID = result
			}
			result, found = subExSingleFinder(line, reSAK, "sak")
			if found {
				p.AWSSecretAccessKey = result
			}
			result, found = subExSingleFinder(line, reST, "st")
			if found {
				p.AWSSessionToken = result
			}
			result, found = subExSingleFinder(line, reRole, "rolearn")
			if found {
				p.RoleArn = result
			}
		}
		profiles = append(profiles, p)
	}
	return profiles, err
}

// Profile is an object to hold a parsed profile
// from an aws creds file
type Profile struct {
	ProfileName        string
	Comments           []string
	Output             string
	Region             string
	AWSAccessKeyID     string
	AWSSecretAccessKey string
	AWSSessionToken    string
	RoleArn            string
}

func main() {
	profiles, err := loadProfileFile("zSampleInput.txt")
	if err != nil {
		panic(err)
	}
	b, err := json.MarshalIndent(profiles, "", "    ")
	if err != nil {
		panic(err)
	}
	fmt.Println(string(b))

}

zSampleInput.txt

[default]
output = json
region = us-east-1
aws_access_key_id = BKIAIABOIDNONBVL987
aws_secret_access_key = 78GisADFEk+knqq2ceGRYRTEERVVIumv7ZW


[dis-other-boi]
# This is that other account
output = json
region = us-east-1
aws_access_key_id = ASIAIOIOVNOIWUEUFYA
aws_secret_access_key = vJVEREWB4Kaw/5Pl0DKWho+g1EfEFFDI0IOnw
aws_session_token = FQoDYXdzEDcaDPwyDDGStuhYrh6KBSLwAVJnpsF+36G9r2PCfT5KZNjVfLw7qXbnd6c1U0ZsV5+3rni/KSW8/5gW2ZkdN1gkO0ERnevergonnagiveyouupRoeqo4K7ycMlqibKo07An+fXjNx4wkTMTrceG3cE8zGeTL5LRJQh/GkoHFFFzD0DSuyFZJjXz56idF+HTi89yQH9NVCLvZz8KVyluvxObjfQW15QE0Yjl+gSFOnsqQe3iCPrOMWYkyyqPFDOtypsFcGG0KlreuycuGQPEub10n0xcu7kajotoRhQBZ3jkmO2Wk2RCjb+a/SBQ==

[dat-old-account]
# DO NOT EDIT
# GOSSAMER MANAGED SECTION
# (Will be overwritten regularly)
####################################################
# ASSUMED ROLE: arn:aws:iam::790812376562953:role/cs/p-operations
# ASSUMED FROM INSTANCE ROLE: NA
# GENERATED: 2018-01-02 16:42:52.513969397 -0500 EST
# EXPIRES@2018-01-02 22:42:51 +0000 UTC
output = json
region = us-west-2
aws_access_key_id = ASIAI6TATMY2YCE65J5Q
aws_secret_access_key = 82HwcMyftGhZxf+KM6HHuaaVTzRsqTPyNi1OU62d
aws_session_token = FQoDYXdzEDcaDINKwUQ4q6ts9XWIGiLwARg+RZM4hzkf59giN5sufa/K9+TOvk6TufnKXBeu467sWQ2BI3xmK1SBrvn2ufundcFud2LtyogO0lGyv8HJfJMc+BFEH/mzhPjudf0wYjGO+d0lOEj2knwnevergonnaletyoudown7pArq/WorFu0om24+gyFgXgc3fzkSVAE/WxAK3gGa18V8Uljj3ZL8JLBRYvMHrDTxw+WbQJ35NBMQ0o2/5DfBiD25znIuYmDXXqPLMmSN7WRiH2QNaPIzG6OwKbPrByvyQyaRMzKZAvGxbiu/syjb+a/SBQ==

[dat-new-account]
# DO NOT EDIT
# GOSSAMER MANAGED SECTION
# (Will be overwritten regularly)
####################################################
# ASSUMED ROLE: arn:aws:iam::7815123567953:role/cs/p-operations
# ASSUMED FROM INSTANCE ROLE: NA
# GENERATED: 2018-01-02 16:46:40.704783754 -0500 EST
# EXPIRES@2018-01-02 22:46:40 +0000 UTC
output = json
region = us-east-2
aws_access_key_id = ASIAAFEFVWEFQWXS4KVQ
aws_secret_access_key = MepbtEPH4EwLEopopamvoieDYJtKjruVU6YbNc
aws_session_token = FQoDYXdzEDcaDO5bzx2bPd8bCTdMAiLwAdjWobiKpismc5ybVEhSLbe1b0ADV6TSAugnRB7LrrSTT0TNIIAD1LBUC7swTHsMo0wCiwvLVynLaBefDEjQsRyi/IW7niJnallbymyself4K+PJIl912G+YNMDSoYMlyqd885qhYUJMH7G04DPKZWL0x8v0hr2gk73XaIOxcAAIkz12TEjk5ro4Y1f4wjcqUFO15CVM0CXiyYNTcxVZ4/Nm/kSHYCT9CSM67NULGf2jm+Ng9xijA+6/SBQ==

zSampleOutput.json

[
    {
        "ProfileName": "default",
        "Comments": null,
        "Output": "json",
        "Region": "us-east-1",
        "AWSAccessKeyID": "BKIAIABOIDNONBVL987",
        "AWSSecretAccessKey": "78GisADFEk+knqq2ceGRYRTEERVVIumv7ZW",
        "AWSSessionToken": "",
        "RoleArn": ""
    },
    {
        "ProfileName": "dis-other-boi",
        "Comments": [
            "# This is that other account"
        ],
        "Output": "json",
        "Region": "us-east-1",
        "AWSAccessKeyID": "ASIAIOIOVNOIWUEUFYA",
        "AWSSecretAccessKey": "vJVEREWB4Kaw/5Pl0DKWho+g1EfEFFDI0IOnw",
        "AWSSessionToken": "FQoDYXdzEDcaDPwyDDGStuhYrh6KBSLwAVJnpsF+36G9r2PCfT5KZNjVfLw7qXbnd6c1U0ZsV5+3rni/KSW8/5gW2ZkdN1gkO0ERnevergonnagiveyouupRoeqo4K7ycMlqibKo07An+fXjNx4wkTMTrceG3cE8zGeTL5LRJQh/GkoHFFFzD0DSuyFZJjXz56idF+HTi89yQH9NVCLvZz8KVyluvxObjfQW15QE0Yjl+gSFOnsqQe3iCPrOMWYkyyqPFDOtypsFcGG0KlreuycuGQPEub10n0xcu7kajotoRhQBZ3jkmO2Wk2RCjb+a/SBQ==",
        "RoleArn": ""
    },
    {
        "ProfileName": "dat-old-account",
        "Comments": [
            "# DO NOT EDIT",
            "# GOSSAMER MANAGED SECTION",
            "# (Will be overwritten regularly)",
            "####################################################",
            "# ASSUMED ROLE: arn:aws:iam::790812376562953:role/cs/p-operations",
            "# ASSUMED FROM INSTANCE ROLE: NA",
            "# GENERATED: 2018-01-02 16:42:52.513969397 -0500 EST",
            "# EXPIRES@2018-01-02 22:42:51 +0000 UTC"
        ],
        "Output": "json",
        "Region": "us-west-2",
        "AWSAccessKeyID": "ASIAI6TATMY2YCE65J5Q",
        "AWSSecretAccessKey": "82HwcMyftGhZxf+KM6HHuaaVTzRsqTPyNi1OU62d",
        "AWSSessionToken": "FQoDYXdzEDcaDINKwUQ4q6ts9XWIGiLwARg+RZM4hzkf59giN5sufa/K9+TOvk6TufnKXBeu467sWQ2BI3xmK1SBrvn2ufundcFud2LtyogO0lGyv8HJfJMc+BFEH/mzhPjudf0wYjGO+d0lOEj2knwnevergonnaletyoudown7pArq/WorFu0om24+gyFgXgc3fzkSVAE/WxAK3gGa18V8Uljj3ZL8JLBRYvMHrDTxw+WbQJ35NBMQ0o2/5DfBiD25znIuYmDXXqPLMmSN7WRiH2QNaPIzG6OwKbPrByvyQyaRMzKZAvGxbiu/syjb+a/SBQ==",
        "RoleArn": "arn:aws:iam::790812376562953:role/cs/p-operations"
    },
    {
        "ProfileName": "dat-new-account",
        "Comments": [
            "# DO NOT EDIT",
            "# GOSSAMER MANAGED SECTION",
            "# (Will be overwritten regularly)",
            "####################################################",
            "# ASSUMED ROLE: arn:aws:iam::7815123567953:role/cs/p-operations",
            "# ASSUMED FROM INSTANCE ROLE: NA",
            "# GENERATED: 2018-01-02 16:46:40.704783754 -0500 EST",
            "# EXPIRES@2018-01-02 22:46:40 +0000 UTC"
        ],
        "Output": "json",
        "Region": "us-east-2",
        "AWSAccessKeyID": "ASIAAFEFVWEFQWXS4KVQ",
        "AWSSecretAccessKey": "MepbtEPH4EwLEopopamvoieDYJtKjruVU6YbNc",
        "AWSSessionToken": "FQoDYXdzEDcaDO5bzx2bPd8bCTdMAiLwAdjWobiKpismc5ybVEhSLbe1b0ADV6TSAugnRB7LrrSTT0TNIIAD1LBUC7swTHsMo0wCiwvLVynLaBefDEjQsRyi/IW7niJnallbymyself4K+PJIl912G+YNMDSoYMlyqd885qhYUJMH7G04DPKZWL0x8v0hr2gk73XaIOxcAAIkz12TEjk5ro4Y1f4wjcqUFO15CVM0CXiyYNTcxVZ4/Nm/kSHYCT9CSM67NULGf2jm+Ng9xijA+6/SBQ==",
        "RoleArn": "arn:aws:iam::7815123567953:role/cs/p-operations"
    }
]